Content uploaded by Yanlai Wu
Author content
All content in this area was uploaded by Yanlai Wu on Feb 20, 2023
Content may be subject to copyright.
379
“I Am Concerned, But. ..”: Streamers’ Privacy Concerns and
Strategies In Live Streaming Information Disclosure
YANLAI WU∗,University of Central Florida, USA
YAO LI, University of Central Florida, USA
XINNING GUI, Pennsylvania State University, USA
Live streaming is a popular synchronous social media platform that allows users to disclose information to vast
audience in real time. It has been increasingly studied in recent years for its unique functions of disseminating
user-generated content, enriching streamers’ self-presentation, curating online social interactions and fostering
online communities. However, little research has been done to explore the privacy issues in live streaming.
In the present paper, we aim to understand streamers’ privacy concerns and strategies in their information
disclosure on live streaming. From an interview study with 20 streamers, we found that they expressed
concerns and carefully managed their information disclosure based on whether the disclosure would enhance
or weaken their attractiveness to the audience and whether it would disturb their interpersonal boundary
with the audience. They adopted various technical and behavioral privacy management strategies to cope
with their concerns, but encountered a series of usability and cognitive burdens. Based on the ndings, we
present design implications to improve the privacy management on live streaming.
CCS Concepts: •Security and privacy
→
Social aspects of security and privacy;Human and societal
aspects of security and privacy;
Additional Key Words and Phrases: privacy; live streaming; streamer; information disclosure; social media
ACM Reference Format:
Yanlai Wu, Yao Li, and Xinning Gui. 2022. “I Am Concerned, But.. . ”: Streamers’ Privacy Concerns and
Strategies In Live Streaming Information Disclosure. Proc. ACM Hum.-Comput. Interact. 6, CSCW2, Article 379
(November 2022), 31 pages. https://doi.org/10.1145/3555104
1 INTRODUCTION
Live streaming is a unique form of social media that allows users to share content and interact with
their audience directly in real time [
85
]. Due to the improved network bandwidth and widespread
high-denition cameras on various devices, live streaming has become increasingly popular [
19
,
29
,
50
,
54
,
55
]. For example, Twitch, the largest live streaming platform in the US, has 41.5 million
users and viewers spent 6.34 billion hours watching Twitch in the rst quarter of 2021 [
15
]. DouYu,
the largest of its kind in China, owned 163.6 million monthly active users in 2019 [
74
], which was
23.6 million monthly active users more than Twitch during the same period [100].
Streamers adopt various practices to present themselves to the audience, such as displaying
distinct personal characteristics [
30
,
82
], dressing themselves decently [
48
], tailoring streaming con-
tent to the audience [
48
], adding entertaining stories, and maintaining social media content outside
Authors’ addresses: Yanlai Wu, yanlaiwu@knights.ucf.edu, University of Central Florida, 4000 Central Florida Blvd, Orlando,
Florida, USA, 32816; Yao Li, University of Central Florida, 4000 Central Florida Blvd, Orlando, Florida, USA, 32816, yao.li@
ucf.edu; Xinning Gui, Pennsylvania State University, State College, Pennsylvania, USA, 16801.
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee
provided that copies are not made or distributed for prot or commercial advantage and that copies bear this notice and the
full citation on the rst page. Copyrights for components of this work owned by others than the author(s) must be honored.
Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires
prior specic permission and/or a fee. Request permissions from permissions@acm.org.
©2022 Copyright held by the owner/author(s). Publication rights licensed to ACM.
2573-0142/2022/11-ART379 $15.00
https://doi.org/10.1145/3555104
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
379:2 Yanlai Wu, Yao Li, and Xinning Gui
live streaming [
54
,
69
,
85
]. As personal information disclosure is a primary way of self-presentation
[
21
], it is inevitable that streamers would disclose large amounts of personal information when
utilizing these strategies, which could lead to privacy issues. There has been news that reports how
streamers’ real names, home addresses, and identities were accidentally revealed in live streaming,
causing their privacy and security to be threatened by vicious viewers [23,28].
However, few studies have investigated streamers’ privacy concerns and practices in their
live streaming information disclosure. Understanding the privacy issues that streamers have en-
countered is important because the privacy management in live streaming is complicated by the
performance-driven, socioeconomic, real-time, on-camera and public nature of live streaming. This
may lead to unique and complex practices and challenges in privacy management that do not exist
on traditional social media: First, streamers’ information disclosure has socioeconomic implications.
It is part of the content that streamers create, which is critical for them to draw and retain an
audience (e.g., by making viewers feel entertained) and make revenue [
46
,
71
]. Streamers might
struggle between privacy protection and content creation. Second, the information production
and consumption on live streaming happen concurrently, which means that privacy management
strategies that are commonly used on asynchronous social media, such as previewing shared con-
tent and withdrawing unwanted content, may not be successful on live streaming. Third, streamers
are usually on camera instead of presenting themselves as avatars or anonymous proles, while
the audience is usually anonymous [
86
]. Streamers’ identiable information is exposed and could
potentially cause privacy risks, such as stalking and harassment. Lastly, live streaming is mostly
open to the public audience. It may entail challenges for streamers to control the live audience and
gauge who will have access to their disclosed information. In summary, streamers might face new
types of privacy issues and challenges in information disclosure that might not exist on traditional
asynchronous social media. It is important to explore such potential issues and understand how to
design in order to support online streamers’ privacy management.
The present study aims to investigate streamers’ privacy concerns and strategies with infor-
mation disclosure on live streaming platforms, which will untangle the intertwined relationships
among privacy management, content creation, and revenue generation and uncover the privacy
challenges in real-time information as disclosed to the public. Specically, we seek to answer the
following research questions:
RQ1: What privacy concerns do streamers have in live streaming?
RQ2: What are the coping strategies that streamers adopt to manage their information
disclosure in live streaming?
To answer the research questions, we conducted 20 semi-structured interviews with streamers.
We found that streamers struggled between privacy protection and attractiveness maintenance.
Certain information disclosure could make streamers’ performance attractive to the public audience,
but raised signicant privacy issues for streamers, such as over-exposure of sensitive personal
details and disturbance of their boundary with public audience. They thus adopted various privacy
management strategies both technically and behaviorally to mitigate these concerns in order
to make a trade-o between attractiveness and privacy cost. Based on these privacy challenges
reported by our interviewees, we put forward the design implications for live streaming platforms.
Our contributions are three-fold: 1) we provide an empirical account of privacy challenges in a novel
context - live streaming; 2) we investigate streamers’ attitudes and practices in privacy management
during the live performance creation, which advance our understanding of information sharing in
live streaming; 3) we suggest design implications for privacy support to reduce streamer’s privacy
concerns and support their privacy management.
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
“I Am Concerned, But. . . ”: Streamers’ Privacy Concerns and Strategies In Live Streaming Information Disclosure 379:3
2 RELATED WORK
2.1 Privacy on Social Media
Information disclosure describes one’s privacy regulation behavior, namely how one attempts
to control his or her information ow to others. Altman dened privacy as the selective control
of access to the self or one’s group, and self-denition (i.e., what is me and not me) as a central
goal of privacy [
2
]. He argued that privacy regulation served a set of functions to meet people’s
needs in self-denition in the social world: 1) constructing self-identity, including one’s denition
of self, capabilities, and limitations. This function is similar to Westin’s personal autonomy [
93
].
Self-identity is the ultimate goal of privacy regulation because regulating access to the self or others
allows people to develop a sense of individuality. 2) regulating one’s interaction boundaries with
others in the social environment, which was also identied by Westin as “limited and protected
communication” [
93
]. Regulating boundary refers to how people set boundaries between themselves
and others to achieve a desired balance between openness and closeness so that they can obtain
important information about their social environment and construct a social denition of the self. 3)
assessing one’s experience with others and formulating future behaviors, which is close to Westin’s
“self-evaluation” [
93
]. It refers to how people individually review information from interaction
with others so that they can develop reection, assessment, and interpretation of themselves in
relation to others. These three functions explain why people need privacy and what social needs
privacy regulation serves.
Extensive research has extended Altman’s privacy regulation theory in the context of social
media and found many privacy issues arising because the above privacy functions are not functional
and users’ social needs regarding self-denition are not met on social media. First, the potential of
exposing large amounts of personal information [
27
] and the collapse of dierent types of social
contexts [
62
] make it challenging for users to construct their desired self-identity on social media.
Research has shown that users were not able to eectively maintain separate parts of their identity
on social media, such as their professional identity [
83
], cultural identity [
44
], peer identity [
1
,
6
],
political/religious identities [
62
], and gender/sexual/racial identities [
11
], due to the challenges
in managing information ow under dierent social contexts on social media. Second, the lack
of eective audience control techniques makes it dicult for users to regulate their interaction
boundaries with others on social media. For example, several studies reported that users struggled
with the boundaries between personal social networks and professional life [
44
,
67
,
83
], weak ties
[84], unwanted relationships [49,77], and public space [1,59,90] on social media.
Research has reported various strategies that users adopt to manage their social identities and
regulate their boundaries with others on social media. Most prior work on identity management
was built on Erving Goman’s theory which frames identity management as theatrical performance
[
5
]. When interacting with others, people attempt to selectively disclose their information and
highlight the content that can deliver desired impressions, like the performers who selectively
present information and act to the audiences to maintain their frontstage appearances. They also
manage the expression of self-identity by projecting dierent personas depending on the audience
and broader context. For instance, Twitter users target tweets toward their perceived audience’s
interest and deliberately avoid topics that followers might not agree with [
62
]. Facebook users
prefer to post group photos that show one’s social connectedness and popularity [
102
]. Other social
media users also create multiple proles so that they can present themselves in dierent ways
[
61
]. Regarding boundary regulation strategies, Altman outlined several coping mechanisms that
individuals employed in the oine setting to achieve desired interpersonal boundaries, such as
verbal communication (i.e., language styles, voice dynamics, etc.), non-verbal communication (i.e.,
blocking, averting direct eye contact, etc.), and environmental mechanisms (i.e., shutting the door,
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
379:4 Yanlai Wu, Yao Li, and Xinning Gui
keeping distance, etc.) [
2
]. Based on these mechanisms, researchers have reported various boundary
regulation strategies on social media, such as censoring oneself [
16
,
81
], blocking unwanted contacts
[
95
], withdrawing unwanted content [
94
], segregating audience [
37
,
92
], negotiating with others
[14], and compromising [94].
In sum, previous social media privacy literature found various concerns and strategies that users
have when regulating their boundaries and managing their identities on social media. However,
most social media examined are asynchronous, which means that the content production and
consumption are in sequence rather than in real-time. Such ndings may not apply to privacy
management on live streaming platforms where information sharing and consuming are in real-time.
For instance, in asynchronous social media, users have a chance to self-censor, edit, and withdraw
their posts, which is unlikely in real-time information broadcasting. Furthermore, live streaming
is mostly open to public viewers who can join the live streaming and interact with the streamers
in real-time [
30
,
34
]. Thus, it is challenging for streamers to get familiar with and tailor their
information-sharing to the current viewers in real time. Building on the rich ndings on privacy
concerns and management on asynchronous social media, our study aims to extend Computer-
Supported Collaborative Work (CSCW) and Human-Computer Interaction (HCI) research on social
media privacy by untangling live streamers’ privacy concerns and management and exploring
design implications to support their privacy management.
2.2 Related Work on Live Streaming
Previous work on live streaming has shown that performance creation is a central element in live
content generation. Research has shown that streamers not only present the content, but also
perform their desired identities to the audience [
48
]. Attractive performance in streamers’ live
broadcast can include talented performance [
35
,
36
,
57
], demonstrations of skills [
36
,
80
], tours and
sightseeing [
36
,
57
], outdoor activities [
55
], gameplays [
35
,
57
], and other aspects of life. Attractive
performances can bring more audience who might give monetary compensation to the streamers
[
8
,
24
,
99
]. Streamers recognize the value of their performance and can potentially make a living
from the monetary rewards from the audience [
53
,
99
]. Additionally, through creating performance,
streamers can develop their “personal brand” [
85
], become famous [
24
], disseminate knowledge
[54], and build online communities [34].
To better understand the attractiveness in streamers’ live performances, we borrow McCroskey
and McCain’s framework [
63
] about interpersonal attraction. McCroskey and McCain’s work
classies interpersonal attraction into three dimensions: physical, task, and social attraction [
63
].
Physical attraction is determined by a person’s physical features. The more physically attractive
one is, the more sociometrically popular. Task attraction refers to our perceptions of whether a
person can assist us to get what we desire. People tend to be attracted to those who are capable of
and willing to fulll their needs and achieve their goals. Social attraction is the socioemotional
dimension of interpersonal attraction, which reects a social or personal liking property. In other
words, social attractiveness implies whether a person is pleasant to be with and could potentially
become a friend.
Self-presentation is a common and important strategy for streamers to maintain attractiveness
in live content broadcasting. Several studies have investigated streamers’ self-presentation strate-
gies in live streaming. Generally, streamers would be on camera, adjust the aesthetics (dressing,
hairstyle, and appearance), decorate physical functions (vocal and physical), research on background
information/knowledge to prepare for their performance, share personal stories and opinions, add
entertaining commentary and casual chat, and highlight distinct personal characteristics, styles and
personalities [
48
]. Another study on female and LGBTQ streamers similarly reports that streamers
develop two core strategies to present themselves through performance, including manipulating
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
“I Am Concerned, But. . . ”: Streamers’ Privacy Concerns and Strategies In Live Streaming Information Disclosure 379:5
webcams and microphones to highlight face and voice and purposely managing clothing and
appearance [
25
]. Streamers also utilize social media outside the live streaming platforms to present
themselves to viewers, such as sharing their social media proles, posting about their skills and
sending pre-recorded videos to audience through direct messages [54].
Since self-presentation involves large amounts of personal information disclosure, potential
privacy issues might arise in streamers’ performance. However, the majority of privacy research in
the context of live streaming has focused on bystanders’ privacy concerns, namely when people
are streamed by others. When streams occur in public spaces, bystanders can be easily captured in
live videos [
55
], thus having concerns about identities being exposed to an unknown audience [
78
].
Another study on video game live streaming also reported bystanders’ concerns with accidentally
disclosing personal information and causing unwanted consequences [
50
]. Thus, bystanders adopt
strategies such as withholding sensitive information and collaborating/negotiating with the stream-
ers to protect their personal information from being streamed [
50
]. To solve the privacy issues of
bystanders, researchers have proposed to obscure details of privacy-sensitive objects in live videos
[
31
]. While these prior studies focused on bystanders in live streaming, few studies have discussed
streamers’ privacy issues. One study briey described the possibility that outdoor live streamers
could accidentally disclose their passport numbers, credit card numbers, and vehicle license plates
while streaming [
55
]. Another study indicates that there is a special group of streamers called
“Nakanohitos” who dub for VTubers, and tend to not disclose their personal identity to the public
due to privacy concerns and corporations’ policies [56].
Other aspects in the live streaming literature cover viewer’s watching experience, live content
moderation and live streaming community fostering. Prior work has shown that the viewers watch
live streams to relax [
57
,
79
], kill time [
43
,
57
], keep up with “fashion” [
52
], make more friends
[
32
], communicate with others [
47
,
57
,
79
], improve skills [
13
,
26
,
79
], nd a community [
12
,
30
],
gain new knowledge [
53
,
54
], or seek the feeling of company [
57
]. To motivate viewers to keep
coming back to the live streams, streamers commonly make eorts to interact with the viewers and
maintain active online communities [
79
]. Streamers’ personalities, attitudes, and values decide who
will stay in their communities [
30
,
79
]. In order to reduce inappropriate content in live streaming,
streamers adopt various technical and behavioral strategies to moderate the live chat in their
streams [
9
,
43
,
97
,
98
]. For example, streamers adjust the pace of the game to meet audience’s
expectations [
43
]. Streamers on Twitch take advantage of moderation tools such as chat control,
viewer control, content control, and setting control to manage their channels and better interact
with their audience [9].
In sum, prior work shows that streamers are prone to present attractive performances to their
audience, which could generate nancial revenues for streamers. Thus, streamers adopt various
self-presentation strategies so that their performance attracts, satises, and entertains the audience.
However, little attention has been focused on streamers’ privacy concerns and practices in live
performance creation. As streamers disclose large amounts of personal information in producing live
performances, their privacy concerns and strategies might interplay with their desire to generate
an attractive performance and to seek nancial revenues. They might encounter tensions between
content creation and privacy protection. Additionally, while viewers mostly remain anonymous,
streamers share a great deal of identiable information in live streaming, such as their physical
appearance and physical locations, which could potentially cause privacy risks, such as boundary
disruption, stalking and harassment. Thus, investigating the privacy issues in streamers’ content
creation is of great importance. Our work will ll the gap by studying streamers’ concerns with
information disclosure and their coping strategies to manage these privacy concerns.
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
379:6 Yanlai Wu, Yao Li, and Xinning Gui
3 METHODOLOGY
3.1 Study Site
We conducted 20 semi-structured interviews with DouYu streamers in October 2020. We chose
DouYu as the study site because: 1) DouYu is one of the most popular live streaming platforms
and has been studied several times in prior work [
101
,
103
]; 2) DouYu shares many privacy and
information disclosure features with other popular live streaming platforms (see Table 1 for a
detailed comparison). Similar to other live streaming platforms, streamers on DouYu are mostly
equipped with a webcam, a mic, and a computer, and/or a mobile phone to run the streaming
platform. Streamers are allowed to: 1) enable webcam anytime to share their video images; 2)
enable/disable their video/audio sharing, start/pause streaming, and enable/disable screen sharing
at any time; 3) apply “beauty lter” to enhance their images in the video, such as smoothing the
skin, tuning the skin tones, adjusting the size of the image, and adding lters or other eects; 4)
enable the “mirror my video" feature when they want to check their looks in the live video from the
audience view; 5) set up a “virtual background”, which is to display an image as the background in
the live video and thus hide the streamers’ real background; 6) add an “overlay”, namely inserting
an image or texture to hide the information on the screen sharing that they want to block from the
audience; 7) control the specic content to be displayed on screen and audio sharing (also called
“manage sources of streaming content”), such as which screen, software, webpage, and screen
zones to be live broadcasted to the audience, and which sound of the microphone, background
music and audio chat to be delivered in the audio sharing. These seven features are listed in Table
1. Both overlay and source management are eective for managing partial information disclosure.
If streamers want to restrict the entire screensharing, the most convenient way is to pause stream,
which freezes the entire video broadcasting. However, this feature does not pause the ongoing
audio sharing. As all these features can also be found on other popular live streaming platforms,
we thus concluded that the ndings based on DouYu are generalizable.
The channels on DouYu are mostly open to the public audience. Audiences can watch streams
either in live streaming mobile apps or on websites, with the streamers’ video image, screen sharing,
and audio at the same time. The live chat box is where streamers and audiences communicate
in real-time. The content posted in the live chat is called Danmaku and ows on the screen [
55
].
Apart from the interactive live chat, streamers can also receive virtual gifts from their audience.
Virtual gifts, worth up to dierent values from cents to hundreds of dollars, are shown as animated
cartoons on the live stream interface and are visible to the audience [
57
]. In sum, DouYu is a popular
live streaming platform and shares many disclosure management features with other popular live
streaming platforms. We thus chose it as our study site.
# Features DouYu Huya YouTube Live Facebook Live Twitch
1 Enable webcam Yes Yes Yes Yes Yes
2 Video/audio sharing Yes Yes Yes Yes Yes
Pause stream Yes Yes Yes Yes No
Screenshare Yes Yes Yes Yes Yes
3 Beauty lter Yes Yes Yes Yes Yes
4 Mirror my video Yes Yes Yes Yes Yes
5 Virtual background Yes Yes Yes Yes Yes
6 Add overlay Yes Yes Yes Yes Yes
7 Manage sources of streaming content Yes Yes Yes Yes Yes
Table 1. Comparison of disclosure-related features between DouYu and other popular live streaming platforms
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
“I Am Concerned, But. . . ”: Streamers’ Privacy Concerns and Strategies In Live Streaming Information Disclosure 379:7
# Age Gender Occupation Length of streaming # Viewers Topics
1 27 Male Student 5 times 5+ Mahjong
2 28 Female Student 3-4 years 300+
Warcraft,
PlayerUnknown’s Battlegrounds,
Call of Duty
3 23 Male Student 15 days 30+ League of Legends
PlayerUnknown’s Battlegrounds
4 28 Male Civil Servant 1 year 20+ JX3 (game)
5 20 Female Full-time Streamer 5 years 600+ Singing, Chatting
Casual games
6 22 Male Freelancer 1 year 200+ Dota
7 26 Male Teacher 3 months 5+ Hearthstone
8 25 Male Student 2 years 80+ Hearthstone
9 20 Female Student 3 years 20,000+ Overwatch,
PlayerUnknown’s Battlegrounds
10 22 Male Student Half year 300+ Honor of Kings
11 22 Female Student Half year 50+ Singing, Chatting
12 18 Male Crew 3-4 months 10,000+ Honor of Kings
13 27 Female Full-time Streamer 2 months 10+ Singing, Chatting
14 28 Male Software Engineer 1 month 8+ Singing, Gwent: The Witcher Card Game
15 23 Male Student 3 months 1000+ League of Legends
16 20 Male Student 1 month 70,000+ PlayerUnknown’s Battlegrounds,
League of Legends
17 30 Male Entrepreneur Half year 5+ Warcraft, PlayerUnknown’s Battlegrounds,
League of Legends
18 22 Female Student 16 days 1000+ Chatting
19 29 Male Financial Analyst 1 year 10-20+ Hearthstone
20 23 Female Student 2 months 10,000+ Apex Legends
21 29 Male Computer Scientist 1.5 years 500+ Algorithm Teaching
22 27 Male Teacher 2 years 700+ Casual Games, Chatting, Outdoor Activities
23 22 Female Student 1 year 200+ Casual Games
24 20 Female Full-time Streamer 2 years 1000+ Casual Games
25 26 Female Full-time Streamer 1 year 3000+ Singing, Chatting, Casual Games
Table 2. Demographics of Interviewees
3.2 Data Collection
The 20 interviewees were recruited through three methods: 1) posting of recruitment information in
the university WeChat groups (the most widely adopted social media in China), through which four
participants were recruited; 2) posting of recruitment ads on Weibo (the most popular microblogging
website in China) under the DouYu-related topics, from which three participants were recruited;
3) thirteen additional participants were recruited through snowball sampling by the referrals of
the author’s personal social network and streamers of those who have been interviewed. All the
recruited streamers were over 18-years of age and had experience in live streaming (see Table 2 for
the interviewees’ demographic information). We carefully considered full-time/part-time, length
of streaming and number of viewers when recruiting participants because we wanted to explore
privacy concerns and strategies for dierent types of streamers. The participants we recruited
included both full-time and part-time streamers, those who have been streaming for a few days to
several years, and streamers who have just a few viewers to those who have tens of thousands.
The interviews were conducted through WeChat audio chat in Mandarin. Each interviewee was
compensated 70 (about 10 dollars) which was sent through WeChat Red Pocket (a money transfer
function on WeChat). The IRB of this interview study was approved by the university and each
interviewee was informed of the research purpose and procedure prior to being interviewed.
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
379:8 Yanlai Wu, Yao Li, and Xinning Gui
Each of our semi-structured interviews took between 30 to 60 minutes. Each interview started
with some general questions to build rapport with participants, such as the length of their live
streaming experience, the content they used to stream, and their motivations to stream. We then
asked about their information sharing during live streaming. We put emphasis on probing in
regard to what information they liked to share, what information they preferred not to share,
their experience and concerns of information disclosure, and their strategies on managing infor-
mation disclosure. The specic interview questions can be found in Appendix A. The interviews
were recorded for transcription purposes, with consent from each interviewee. After the authors
transcribed the interviews and anonymized identiable details, the audio recordings were deleted.
3.3 Data Analysis
In conducting our research we applied thematic analysis [
7
] to evaluate our interview data using
an inductive approach [
58
]. It’s important to note that all the researchers who participated in data
analysis were familiar with live streaming and privacy research. Our process began by scrutinizing
the data then individually noting initial ideas related to information disclosed. Next, we compared
our initial data codes and combined them into a single list through rounds of discussions leading
to a total of 89 codes. According to the initial coding, we advanced our analysis to gather the
overarching themes from our codes in extensive rounds of discussion. We also went back and
forth between the generated themes and the dataset to collate codes into potential themes and
sub-themes. The next step was to rene the themes and sub-themes to ensure internal homogeneity
and external heterogeneity [
58
]. The nal thematic map consists of three primary themes: privacy
concerns in live streaming, privacy strategies in live streaming, and challenges streamers have
encountered in privacy management. All the quotes from our participants were translated into
English when reporting the data in the paper. To protect our participants’ privacy, we use numbers
to denote dierent interviewees.
4 STREAMERS’ PRIVACY CONCERNS
Generally, we nd two categories of privacy concerns that participants had experienced during
their information disclosure in live streaming. First, streamers were concerned that disclosing or
withdrawing certain personal information might undermine their attractiveness to the audience.
Second, streamers had concerns in regard to their interpersonal boundary with the audience. We
elaborate on these concerns in the next two sub-sections.
4.1 Tensions Between Privacy and Aractiveness
Our participants reported two major tensions between privacy and attractiveness in live streaming.
First, while they were worried that disclosure of certain information, such as physical features, body
movement skills, and personal preferences, would reveal too much personal information to the
audience, they found such information made their live performance more attractive, thus decided
to compromise their privacy in exchange for attractiveness. Second, certain personal information,
though authentic about streamers, raised concerns to streamers because authentic information
would unfavorably impact their attractiveness.
Previous research has found that perceived media character attractiveness inuences viewer
responses to media personae, thus playing a key role in parasocial interactions [
39
,
40
,
73
]. In our
study, our streamer participants, as media personae themselves, acknowledged the importance of
demonstrating their attractiveness as a way to attract and retain viewers, while at the same time
struggled with the tensions between privacy and attractiveness.
For our participants, there were complex tensions between preserving private information and
constructing such attractiveness. They were aware that disclosing certain personal information
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
“I Am Concerned, But. . . ”: Streamers’ Privacy Concerns and Strategies In Live Streaming Information Disclosure 379:9
would attract audience, which would lead to more monetary income (as most live streaming
platforms allow audiences to give monetary gifts to the streamers) and more entertaining content.
For instance, P10, a male 22-year-old student who had streamed for six months on DouYu with
more than 300 viewers (shortened as “male, 22, student, half year, 300+”), reported, “If I have a lot
of audiences, one day I make an impressive kill (in the game) and they (audience) are likely to send me
gifts. Then my ranking (on DouYu) will go up. It means more people will be attracted to my channel.
There will be more chat, more interaction, more subscriptions, and of course more gifts.”
However, sharing such personal information oftentimes contradicted their privacy attitudes
and concerns. They had to make a trade-o between privacy and attractiveness. In this section, we
report three types of tensions between streamers’ privacy and attractiveness, inspired by previous
work’s classications of interpersonal attractiveness [40,42,63], in the next three subsections.
4.1.1 Tradeos Between Privacy and Physical Aractiveness. Physical attractiveness refers to the
appealingness of a person’s outer appearance [
40
,
42
,
63
]. Six participants (P1, P2, P5, P10, P14, P15)
ranging from new streamers to experienced streamers explicitly explained to us that they were
concerned with disclosing information about their physical features, such as looks, appearance,
apparel, gures, voice and actions, during live streaming. They felt such disclosure exposed too much
personal information to the audience. Other participants, especially most of the female streamers
(P9, P11, P18, P20) mentioned that they hesitated to share such information as well. However, after
failing to attract audience and observing other successful streamers, all those participants realized
that such information could show their physical attractiveness to the audience and thus attract
more viewers and increase their nancial revenues. Therefore, most of them eventually decided to
disclose their physical appearance and even enhance their appearance (e.g., by using makeups). For
example, P5 (female, 20, full-time streamer, 5 years, 600+), a female streamer, described:
I don’t really want to be on the webcam. I mean who wants to? I feel like I have nowhere to hide
(when the webcam is on). Everyone is staring at me. I don’t like that. I didn’t turn on my webcam at
rst. Then there was no viewer at all. So I thought about it. You know, those people on TV show their
faces all the time. Of course, they have pretty face. But it’s about interaction, right? If you treat it (live
streaming) like a profession, showing your face means you respect your audience.
Even when they decided to trade privacy for physical attractiveness, participants were worried
that their real physical appearance was unsatisfying to audiences, which might negatively impact
their attractiveness. They observed that audiences would give more monetary gifts, attention, and
comments to streamers who were better looking and treated less good-looking streamers with less
enthusiasm. Thus, they put eort into improving their physical attractiveness. As P5 continued her
story:
I remember when I turned on my webcam for the rst time, I was looking like a mess. I just hated
makeups and dresses. After like three years, I found that I couldn’t compete with other streamers. Other
streamers were pretty and got tons of gifts, followers and Danmakus. I was like really lame. My viewers
treated me like a random person. You know, people would be nice to you if you got a pretty face. I am a
girl. I don’t look ugly. Why should I give up? So I began to dress up before live streaming.
In these two quotes, P5 articulated how and why she gave up privacy in exchange for physical
attractiveness. As a streamer, she felt that she had no choice but to trade privacy for attractiveness
in order to compete with other streamers regarding the size of audience and nancial revenues.
In addition, most of our participants, including most of the female streamers (P2, P5, P9, P11,
P18, P20) perceived that disclosing satisfying physical features in live streaming was especially
necessary for female streamers, based on their experiences and observations. According to P2
(female, 28, student, 3-4 years, 300+), another female streamer:
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
379:10 Yanlai Wu, Yao Li, and Xinning Gui
I don’t want to turn on webcam. But it is very necessary for girls. Girls get more audience if they
are on camera. You know Xiao TuanTuan (a famous female streamer on DouYu)? She used to attract
audiences with her voice and humor. But she is even on camera now... I remember when I rst streamed,
I only displayed my hands and keyboard (not sharing face in the webcam). Then audiences commented,
“why does she hide her face” “Is it because she looks ugly.” You know. They are just coming for a pretty
face.
P2 emphasized the importance of disclosing looks for female streamers in order to attract a
good size of audience, even though such information sharing revealed more personal details than
disclosing other types of physical features. When she was only willing to share her voice and hands
in live streaming, she often encountered unfriendly audiences which made her feel frustrated. Such
situations rarely happened to our male participants, as P17 (male, 30, entrepreneur, half year, 5+),
a male streamer, told us, “Few male streamers use webcam in live streaming unless they need to do
some singing, dancing and outdoors. There is a big dierence between male and female streamers in
using webcams. Female streamers have to do that.”
4.1.2 Tradeos Between Privacy and Task Aractiveness. Task attractiveness refers to a person’s
success in accomplishing tasks [
40
,
63
] desired by the audience. People tend to be attracted to those
who are capable of and willing to fulll their needs and achieve their goals [
63
]. To achieve task
attractiveness, some of our participants chose to disclose certain personal information to prove
that they were capable of doing the tasks they streamed. For instance, P14 disclosed his educational
background in psychology to demonstrate that he was better at solving mental health related issues
than other streamers. Many audiences often came to his live streams and asked for advice: “They
called me the professor because I know lots of stu. I told them I had a background in psychology and
that delivered them a message – I have expertise and I am more professional than other streamers."
Other times, streamers performed certain tasks only for a particular audience, and had privacy
concerns that their performances were observed by a dierent audience. For instance, the audience
sometimes would request streamers to perform certain types of tasks, such as doing some sexually
suggestive actions, which streamers did not feel comfortable doing because of privacy concerns.
Several female streamers told us that they were unwilling to disclose their gender, body movement
skills (i.e., twisting, dancing and bouncing), intimate personal details, and personal preferences
in the content created for live performance. They were worried that such information disclosed
might cause discriminatory harassment from the audience to them and enhance people’s existing
stereotypes and biases towards female streamers. However, they were aware that the audience
expected to see these types of tasks in their streams. Some of them eventually compromised as
such disclosure was eective to enhance task attractiveness and fulll the audience’s needs. Such
scenarios were more frequently brought up by female streamers, especially those who had only
streamed for a short period of time and were desperate for more audience members. For example,
P18 (female, 22, student, 16 days, 1000+), a female streamer who just started to stream on casual
talk but had a growing number of viewers, told us:
They (audience) are here for fun. If you do something sexual but not across the line, like dancing a
little and twisting your body, they will be very happy. Do I reject it? Well, it’s kinda a show, right? You
give them what they want. They won’t ask for too much. Of course, if you are a famous streamer, you
don’t have to do that.. . I am concerned if my friends see me doing that in the streams. I know they
watch streams a lot. That will be so embarrassing if they see me there doing those sexual things. You
know people already have bias towards female streamers.
4.1.3 Tensions Between Privacy and Social Aractiveness. Social attractiveness implies whether
a person is pleasant to be with and could potentially become a friend [
63
]. Participants reported
tension between protecting private information in regard to their social identities and maintaining
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
“I Am Concerned, But. . . ”: Streamers’ Privacy Concerns and Strategies In Live Streaming Information Disclosure 379:11
social attractiveness. Certain personal details about streamers’ social identities, such as real-world
relationships, social life, interests, and opinions, could make streamers socially attractive and likable,
increasing audience’s interest to socialize, interact and gift with the streamers. However, revealing
such personal details might cause unwanted privacy risks to streamers, such as online stalking
and bullying. Thus, the streamers had to balance between over-exposure and over-restriction of
personal information in the online interaction with an audience to make themselves more socially
attractive.
In live streaming, one important way that participants maintained their social attractiveness
was to share personal details, including relationships, everyday life, interests, personalities, and
opinions. Three participants treated the audience as their friends, thus feeling no point in hiding
personal details from the audience, as they believed being honest was important when interacting
with friends. For example, P12 (male, 18, crew, 3-4 months, 10,000+) explained “I love sharing
everything with my audience. They are my friends. I want to be honest with them. They liked to ask
about my relationship, like where we’ve been and what we eat. That’s what friends talk about.”
However, the majority of female participants were concerned with sharing too many personal
details with the audience for social attractiveness, especially details about their real-world social
relationships. Three female participants expressed concerns about disclosing relationship informa-
tion in exchange for social attractiveness because it might cause unwanted consequences, such as
online stalking and bullying. These participants had dierent lengths of streaming experience and
numbers of viewers. For example, P2 (female, 28, student, 3-4 years, 300+) shared:
They would ask me a lot of questions about my relationships, like when I broke up and what type
of guys I am interested in. I usually make up the answers. I won’t tell them the truth.. . They want to
know me better and get close. So they come to my channel more often. But I’m concerned that some of
them might get too close, like they feel they have a good relationship with me and ask for my phone
number and maybe, you know, stalk me.
In this case, P2 was concerned that disclosing relationship information might cause online
stalking. However, such information disclosure could potentially improve her social attractiveness
and strengthen her relationship with the audience. Thus, the streamer decided to use a fake response
to both maintain her social attractiveness and preserve her privacy.
In addition, participants mentioned the tension between disclosing their authentic social identity
and maintaining social attractiveness. Though participants have the desire to be authentic and
honest to their audience, they found not all personal details are likable to the audience, such as
romantic relationships, living environment and private interests/behaviors (i.e., sexy wallpaper).
This information would harm their social attractiveness because it might not be socially desirable.
Therefore, they would restrict such information disclosure or even lie about it. For instance, seven
participants reported that audiences did not like interacting with or sending gifts to streamers who
had already been in a romantic relationship. Hence, these streamers pretended to be single in live
streaming. Such situation was more common to female (n=5) than male participants (n=2). For
example, P20 (female, 23, student, 2 months, 10,000+) said:
The audience knew I had a boyfriend when I streamed a long time ago. At that time, my boyfriend
and I played the League of Legends and we used to sit together when playing and streaming. They called
my boyfriend ‘mouse’ because he sometimes helped me play the game when I went to the bathroom. It
was really weird, they liked to joke around my boyfriend. Then this time, I got some audience was like
‘if you have a boyfriend, I am not coming’. So I didn’t tell them about my boyfriend. I think they want
me to be single. They are not like hoping to develop a romantic relationship with me. But they want to
keep some imagination there, you know, like those have for idols.
When P20 frankly shared her relationship with her audience in her previous live streaming, she
felt the unfriendly attitude from the audience towards her boyfriend, as her audience made fun of
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
379:12 Yanlai Wu, Yao Li, and Xinning Gui
her boyfriend. Thus, she decided to hide her relationship and to perform a “single girl” identity in
her current live streaming to increase the audience’s desire to socialize with and gift her. Based
on her experiences, P20 believed that female streamers risked losing social attractiveness if they
disclosed their romantic partners because audiences expected the female streamers to be single. In
fact, audience members did not plan to be in an actual romantic relationship with the streamers, as
in most cases it was infeasible. Nevertheless, they wanted to keep some possibility there, where
they could socialize with the female streamers in whichever way they preferred regardless of her
actual relationship status. Thus, the streamer decided to hide her relationship and performed as a
“single girl” in her live streaming. This indicated that streamers had to withhold certain information
disclosure and perform their social identity in order to maintain their social attractiveness.
Although participants tried hard to hide undesirable information from harming their social
attractiveness, such information could be accidentally disclosed by their bystanders, such as their
families. Two participants, one experienced and one beginner streamer, had encountered such
issues. For example, P7 (male, 26, teacher, 3 months, 5+) said:
My parents would sometimes ask me to clean my room, or talk to me about some family matters,
while I’m streaming. It’s like if you got a friend visit you, and your mother asked you to clean your
room. It was just so embarrassing.
In this case, the information that would negatively aect the participant’s social attractiveness
was disclosed indirectly by his parents, who appeared in live streaming unexpectedly. The disclosure
was unintended, as it was not the parents’ intention to broadcast such information to the audience.
But it raised privacy concerns to the streamer knowing they could not fully control undesirable
information on their own eorts. Even though they tried hard to perform a socially acceptable
identity to audience members, bystanders’ information disclosed could easily cast a negative impact.
4.2 Tensions between Privacy and Interpersonal Boundary with the Audience
Participants expressed their concerns with interpersonal boundary regulation with their audience.
Generally, they felt frustrated with unwanted physical and virtual contact, harassment and interrup-
tion caused by their accidental information disclosure to the audience. They were also concerned
about accidentally disclosing bystanders’ information to the audience, potentially bringing trouble
to those bystanders.
Although streamers oftentimes traded their privacy for performance attractiveness, they did set
both physical and virtual boundaries between themselves and audience. Physical boundary focuses
on the contact in real world whereas virtual boundary denotes the contact in the cyberspace, such
as sending friend requests, emails and calls. According to Altman [
2
], regulating one’s boundaries
with others is a critical function of privacy management. Our participants described their concerns
with unwanted physical or virtual contact from the audience. Such unwanted contact mostly
involved certain identiable information about the streamers. Such identiable information would
raise signicant privacy concerns to streamers, especially the interpersonal boundary turbulence.
But the information might not be critical in creating attractive performance to audience. Thus,
participants weighed it over attractiveness and set boundaries accordingly. Regulating interpersonal
boundaries between the streamers and their audience was challenging because of the real-time
nature and openness to the public audience on live streaming. We describe participants concerns
with boundary regulation in the next two subsections.
4.2.1 Privacy Concerns about Physical Boundary. Participants described their concerns with unde-
sired physical contact from audiences. Such concern was mostly a result of live streamed disclosure
of streamers’ mailing address, school/workplace address, and whereabouts. Disclose of this infor-
mation was usually not a part of attractive live performance. In many cases, the audience would
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
“I Am Concerned, But. . . ”: Streamers’ Privacy Concerns and Strategies In Live Streaming Information Disclosure 379:13
randomly ask streamers about their hometown, cities they lived in, and schools. For instance,
two participants were asked by the audience to share their mailing addresses with the excuse to
mail gifts to the streamers. Most participants chose to decline such questions and restrict such
information sharing to protect their own privacy.
In addition, streamers were cautious that audiences could sometimes infer location information
from the scenery and landmarks outside the window of the streamer’s room. For example, P13
told us a story he read from the news that a streamer shared the scene outside the window when
streaming. As a result, an audience member found the streamer’s place.
Participants also told us that unwanted physical contact could cause a security threat to the
streamers and streamers’ families, as well as harassment, i.e., audience sent funeral gifts to troll the
streamers (according to P12). What was more bothering was the cost and burden for a streamer
to change address. This indicates that streamers were concerned with information exposure that
would harm their physical boundary and bring them physical security threats. This concern was
mentioned by 10 participants with various lengths of streaming (ranging from 2 months to 3-4
years) and numbers of viewers (ranging from 5+ to 10,000+). For example, P20 (female, 23, student,
2 months, 10,000+) told us:
My audience oered to order food for me. I declined. It’s (address) private. I stream at home.
Some streamers work at the company, so their addresses are mostly the company addresses. But I am
concerned. It’s scary. I know I am not that famous. But you know, you have to change your address (if
exposed). That’s burdensome.
In this case, P20 was concerned about the privacy of her home address. She did not want the
audience to know her exact location in the real world. She also mentioned the unacceptable cost
of addressing such a privacy breach, which made her even more careful to restrict this personal
information.
4.2.2 Privacy Concerns about Virtual Boundary. Unwanted contact happened not only in the
physical world but also in cyberspace. Eight participants, who have dierent streaming experiences
ranging from one year to only a few days and dierent numbers of viewers ranging from thousands
to only a few, described their concern about unwanted virtual contact when accidentally sharing
contact information during live streaming. Such contact information included the phone number,
QQ ID, WeChat ID, in-game ID, and email addresses (WeChat and QQ are two major social media
messengers in China). Most participants disclosed their contact information accidentally, such as
when they opened a document, a webpage, a push-up notication, or a message, or when they
logged in/out of websites and applications during their live stream. As these were normal and
frequent operations in daily life, participants found it dicult to avoid them during a live streaming.
Once the contact information was exposed, streamers could be contacted and harassed by vicious
viewers unexpectedly. For example, P10 (male, 22, student, half year, 300+) told us:
I once saw such accidents in others’ live streaming. It was a game streamer and he’s kinda famous
with around hundreds of thousands of viewers. He was sending some les through QQ and his QQ ID
was exposed. Some audience caught that and all of a sudden, his QQ was exploded, like thousands
of friend requests ew in. I guessed he had to discard that ID. Another streamer exposed his phone
number when he was sharing his phone screen. Again, just in one minute or two, he got tons of calls.
And the stream was disrupted, too. So those could happen to me if I am not cautious enough. It would
destroy my life.
While unexpectedly disclosing one’s own contact information was annoying, participants
worried about accidentally disclosing others’ contact information to the audience, causing them to be
unexpectedly disturbed by remote audience members. For example, a streamer might unintentionally
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
379:14 Yanlai Wu, Yao Li, and Xinning Gui
share a friend’s phone number if called by a friend during a live stream if using a mobile phone. As
P19 (male, 29, nancial analyst, 1 year, 10-20+) said:
Yea, it happened once to me. I was logging in to my friends’ accounts, two accounts actually (upon
my friends’ requests). I thought I stopped sharing the login interface. So I logged in to one account and
opened another window to log in to the second one. But when I was done, I found it (screen sharing)
was not stopped. I was so scared. Fortunately, it was midnight and no one was there (watching). It
would be a disaster if my friends’ accounts got stolen.
When his friends’ account numbers were disclosed, P10 was concerned about a potential threat
directed toward his friends. Disclosing contact information to audiences could harassing phone
calls, emails, messages, and friend requests, as well as “stream sniping”, a practice of playing a
game against a streamer while watching their broadcast to gain an advantage [
82
]. For example, P7
(26, male, teacher, 3 months, 5+) told us:
Sometimes when I open my friend list (in the game), they (audiences) could see my friends there
and my Battle.net ID. I don’t really want them to see that because if I am playing rank (a play mode in
Hearthstone for higher ranking), the audience might want to snipe me in the game based on my ID.
It’s like they are trying to join in the same match with me while watching my streams. And they can
see my cards.
Such “stream sniping” is not uncommon for streamers. There has been much news about how
streamers ran into a set of stream snipers who made trouble to whatever the streamers were playing.
Such virtual unwanted contact was also reported in our interviews.
5 STRATEGIES AND CHALLENGES TO MANAGE PRIVACY IN LIVE STREAMING
To address the concerns in the above section, participants reported various coping strategies they
adopted to manage their information disclosure. Their strategies can be grouped into two categories:
platform-supported strategies and self-developed strategies. However, these strategies were not
awless. Several usability and cognitive problems came in to play when participants applied these
strategies.
5.1 Platform-Supported Strategies
We conclude three types of frequently adopted strategies that are supported by the features
embedded in DouYu: 1) pause streaming; 2) managing sources of streaming content; 3) adding an
overlay. These three features can also be easily found on other live streaming platforms, such as
Twitch, YouTube Live and Facebook Live. DouYu has a button on its interface that allows streamers
to pause the live streaming. Once activated, the live video will freeze. However, the streamer’s
audio is still being broadcasted when this feature is on. DouYu also oers a set of options that allow
streamers to freely manage the source(s) of video/audio feeds. Streamers can select whether they
want to stream videos of the particular application(s) or entire desktop/phone screen, and audios
of the game soundtrack, in-game audio chat, or streamers’ commentary. Lastly, DouYu allows
streamers to add overlays to video content. Streamers can add emojis, pictures, and text as overlays
to cover a portion/zone in their live videos.
Generally, participants reported that they would use these features when they were concerned
that their information disclosure would negatively impact their attractiveness or boundary reg-
ulation, such as when checking emails/messages, making payments, and talking about personal
matters. Streamers who mentioned these strategies had dierent lengths of streaming (from 3-4
years to a few days) and numbers of audience members (from 10+ to 10000+). For example, P2
(female, 28, student, 3-4 years, 300+) mentioned:
I usually broadcast game soundtrack and team chat, because I feel it’s kinda boring if there is
just the BGM (background soundtrack in the games). So I stream the team chat about our strategies
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
“I Am Concerned, But. . . ”: Streamers’ Privacy Concerns and Strategies In Live Streaming Information Disclosure 379:15
and communications in the game. Sometimes we made little jokes in team chat. Those are fun. But
sometimes when we are done the team playing and a few of us left there complaining about some team
members, I will pause my stream because you don’t know who’s watching there. What if they spread it
to others and to the persons we complain about? So I just pause it and might restart it later.
P2 described two situations where she wanted to congure the source of audio disclosure in
her live streaming: 1) when she wanted to improve her task attractiveness, as she believed that
disclosing rich audio content could make her live streaming more interesting to audiences; 2) when
she wanted to prevent her complaint from being heard by others, which might harm her social
attractiveness and boundaries with other players. Managing the audio source was thus eective in
these two situations.
5.2 Self-Developed Strategies
While the live streaming platform oers a set of features to support streamers’ management of
information disclosure, participants still reported their heavy reliance on self-developed strategies
to cope with their privacy concerns. These self-developed strategies include both technical and
behavioral strategies.
Technical strategies are mostly related to software and hardware outside the live streaming
platform. They were adopted by both beginner and experienced participants and by those who
had a considerable number of viewers. For example, participants used webcam lters to make
their physical appearance look better. They disabled the notications in settings of the operation
systems, signed out of the social applications they used, and removed content they preferred not to
share from the desktop and application interface. All these actions were performed on software
systems that were independent of the live streaming platform. They assisted the streamers to
manage information disclosure and block content from the audience. For example, P12 (male, 18,
crew, 3-4 months, 10,000+) told us:
I logged out every possible social media account and disabled all the notications and calls on my
phone (when streaming) just to leave a good impression on the audience. Also, I had some intimate
chats with my girlfriend, and I don’t want them to see that. There was a time, like when I stream for the
rst time or the second, a call came in when I streamed my gameplay. The entire live video just blacked
out, because I only screen shared the game. So when the call came in, the game was interrupted so
there was just a black screen to the audience. I was really panic because my company was advertising
me for a long while on DouYu and I got like ten thousand viewers watching me there. I just whispered
‘I am streaming’ on the call and hang up. I don’t even remember who was on the call cuz I was freaked
out. I lost two to three thousand viewers immediately because they thought my streaming was ended.
In this case, P12 signed out of social media and disabled the notications on his phone because he
wanted to maintain his task attractiveness assuming that receiving calls during live streaming could
drive audiences away. Another motivation for P12 was to protect his and his girlfriend’s privacy.
Interestingly, he mentioned that he also used DouYu’s source management feature, indicating that
streamers combined dierent strategies to manage their information disclosure.
Many hardware devices, such as webcams, microphones, and headsets, also oer functions that
allow streamers to control their information disclosure, such as the on/o button. Participants
noted that they would choose to use such devices during live streaming, as it was quite convenient
to control their information disclosure. Three participants reported that they would use double
devices (either double computer monitors, or one computer monitor plus one mobile device) to
self-censor their physical appearance in live streaming. They would stream on one device and
watch on the other device as an audience. Such hardware-related strategies are obviously outside
of the live streaming platform. For example, P20 (female, 23, student, 2 months, 10,000+) told us:
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
379:16 Yanlai Wu, Yao Li, and Xinning Gui
Because I use webcam, I had to switch the screen from time to time to check how I looked like, am I
too ugly. You know when you play games, you got nervous and looked like an idiot. So I checked my
video image from time to time.. . But when I was really busying ghting in the game, I could forget to
check.
Asides from technical strategies, participants also heavily rely on improvised behavioral
strategies. Such strategies were dependent on streamers’ mental models of the streaming process.
Seven participants mentioned that they would adopt self-related behavioral strategies, such as
improving their physical appearance and changing their attitude. These participants had between
a few days to 5 years of streaming experience and anywhere from just a few viewers to more than
20,000. For example, P13 (female, 27, full-time streamer, 2months, 10+) explained that she would
dress up and wear make-up before her live streaming so as to look beautiful and energetic to the
audience. P8 (male, 25, student. 2years, 80+) would also adjust the lighting setup when starting to
stream, as a reliable lighting setup made him look professional and enhanced the quality of the
webcam feed. Hence, these strategies would address streamers’ concerns regarding their physical
attractiveness.
As participants reported the impact of environmental information (i.e., messy bedroom and
landmarks outside the streamer’s window) on attractiveness and boundary regulation, they would
adopt environment-related strategies to control their background and surroundings. To avoid a busy
background undermining their social attractiveness and exposing physical location, participants
would use a green screen. The streamers would set up a green curtain as the background and
congure the webcam settings to remove everything green; then the streamer would have a
transparent background in the video, such as having only the upper body displayed over gameplay.
For instance, P16 (male, 20, student, 1 month, 70,000+) told us:
I got a green curtain. It’s like a piece of green cloth. When my background is a mess, my audience
might think I am a mess because my personal life is there. So if I happen to have some audience dislike
me, they will judge me, like your room is a mess, you gotta be a messy person.
Streamers also adopted audience-related strategies, meaning they would control information
disclosed based on the live audience. Generally, participants reported that they would adopt dierent
communication strategies when disclosing personal information to the audience, such as faking
information, distracting the audience’s attention, making a joke to switch topics, and simply
ignoring/declining the audience’s request for streamers’ personal information. These strategies
involved nuancing management of the interpersonal communication with the audience, thus
were highly dependent on the streamers’ skills of communication. Participants mostly adopted
such communication strategies when their authentic disclosure would negatively impact their
attractiveness and boundary. Communication strategies are commonly adopted by participants
who have accumulated certain lengths of streaming experience (from 2 months to 5 years) and
numbers of viewers (from 80+ to 10,000+). For example, P2 (female, 28, student, 3-4 years, 300+)
faked information to hide her social relationships that she did not want to disclose to the audience.
However, she also wanted to maintain her attractiveness to the audience. Thus, she used a bit of
exaggeration in her communication strategies so that audience would keep their interest in the
streamer: “I just made up a little story, you know, exaggerated it or just smiled, made a little joke like
’Ah, don’t get me started.’"
As participants had mentioned how their bystanders destroyed the information disclose that
they had tried to control, they often adopted strategies related to bystanders to prevent disclosure of
private information. Similar to audience-related strategies, the strategies with bystanders were also
mostly about communication. Six participants including both beginner and experienced streamers
described that they would communicate with their bystanders, such as parents, partners, and other
family members, in advance about when, where, and why they needed to stream, so as to prevent
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
“I Am Concerned, But. . . ”: Streamers’ Privacy Concerns and Strategies In Live Streaming Information Disclosure 379:17
bystanders from interrupting the live streaming and undermining the streamers’ attractiveness.
For example, P5 (female, 20, full-time streamer, 5 years, 600+) described:
My little brother is only 9 years old. He doesn’t understand streaming is a job. He thinks I’m just
playing games. So he shows up every day when I stream, like sitting next to me, eating food there, and
coming in and out of my room. The last time I was singing in my stream, and he just came in out of a
sudden and sat there, talking to my audience! I told my audience that I ate at 9. He was like ‘no, you
ate at 8’. He even shared my birth date and year with my audience. I was shocked. So this year, I talked
with him like streaming is a serious job. I raised myself on it. He could only enter my room once per
day. And he couldn’t share his name, age, and address in my stream.
In this case, P5’s younger brother was too young to understand the profession of a streamer, thus
found it fun to interrupt his older sister, P5’s live streaming. His behavior, including intentionally
disclosing P5’s personal information, negatively aected P5’s attractiveness and boundary with the
audience. P5 had to explain the signicance of live streaming to her younger brother to prevent
him from interrupting again.
5.3 Limitations of the Privacy Management Strategies
Although participants adopted various strategies to meet their needs for information disclosure,
these strategies were not awless. We summarized two types of challenges that streamers encoun-
tered when deploying privacy management strategies.
5.3.1 Unpredictability, Uncertainty, and Unexpectedness in Privacy Management Strategies. Although
streamers actively implement privacy management strategies, there are privacy concerns that can-
not be addressed by existing privacy management strategies due to the unpredictability, uncertainty,
and unexpectedness in live streaming. Participants reported that while they adopted privacy man-
agement strategies to withhold unwanted information disclosure in live streaming, there were
times when they could not overcome the privacy challenges through the strategies because they
were not able to precisely predict and control the content in the real-time information sharing. Such
problem mostly happened in the content of screensharing and audio sharing, such as notications,
calls and automatic webpage redirect, which could not be predicted in advance, but could easily
disrupt live streaming, reveal sensitive personal information and cause a negative impact on their
attractiveness and boundary regulation. No eective technical features were provided inside or
outside live streaming platforms to assist streamers to predict the occurrence of this accidental
real-time information disclosure. Thus, streamers were not able to timely implement appropriate
strategies to avoid this unwanted information disclosure. For example, P13 (female, 27, full-time
streamer, 2 months, 10+) told us:
I was going to buy something in the game. Then it just automatically and unexpectedly connected to
my Alipay (a popular payment method in China, similar to PayPal). Boom, all my personal information
was exposed. The audience was like ‘OMG, you are so rich!’ I was so embarrassed. I mean I want them
to give me gifts. If they see I’m rich, they might not do that (give gifts).
The uncertainty in audience’s reaction is also part of the unpredictability in live streaming.
Because audience are anonymous and the live streaming is mostly open to public audience, streamers
failed to predict the real-time response from the audience when adopting audience-related privacy
management strategies, thus making their strategies ineective. For instance, participants described
how their audience-related strategies, such as faking, distracting, joking, and ignoring, failed when
the audience kept asking and trolling the streamers. For example, P12 (male, 18, crew, 3-4 months,
10,000+) said:
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
379:18 Yanlai Wu, Yao Li, and Xinning Gui
I used to nd some excuses to distract the audience when I lost in the game, you know, to hide the
fact that I am not that skilled. But some audiences just didn’t buy it. They would blame me, like ‘it’s
your fault. No excuse!’ posting this directly on the Danmaku.
In addition to the unpredictability and uncertainty in live streaming, technical strategies that are
supported by emerging technologies, such as source management and webcam lter, might fail to
work as expected when streamers use them, thus negatively impacting the streamer’s attractiveness
and boundary. Participants mentioned a set of emerging technical features that need to be further
improved. For example, the source management feature should display a message to the audience
if the selected sources (preferred information disclosure) crashed during the live streaming. The
webcam lter may cause an unbalanced skin tone and face shape. For example, P20 (female, 23,
student, 2 months, 10,000+) said: “Those beauty lters allow you to adjust the shape of your face or
your eyes. But if there’s something in front of your face, or if you move in or out of the lters, your
face becomes weird, like being squashed.”
5.3.2 Cognitive Burdens Brought by the Privacy Management Strategies. Even if the privacy man-
agement strategies work eectively, deploying these strategies brings extra cognitive burdens to
streamers. Self-developed strategies often added signicant cognitive burdens to streamers because
these strategies were not supported by the live streaming platform. Participants had to refer to
software, hardware and behavior outside the live streaming platform to deploy their strategies,
which cost them much eort. This was even more burdensome during live streaming because
streamers were mostly performing multiple tasks at the same time, such as keeping up their
competent performance and interacting with the audience. Thus, participants reported that using
self-developed strategies caused too much cognitive burden on them. For example, P9 (female,
20, student, 3 years, 20,000+) told us that she needed to wear makeup, contact glasses, and fake
eyelashes while streaming games, which made it hard for her to concentrate and it impacted her
performance. P5 (female, 20, full-time streamer, 5 years, 600+) complained that she had to check
the content of screensharing several times every time before she streamed in case inappropriate
content was broadcast, such as WeChat and QQ messages.
Participants also complained about the cognitive burden that they had to continuously remind
themselves to implement privacy management strategies during live streaming. This is because some
information disclosure that is inappropriate in live streaming is actually common and appropriate
in the oine setting, such as sending intimate messages to social relationships and checking emails.
Thus, streamers might sometimes fail to realize that the context of information sharing had changed
during live streaming. They might disclose unwanted information without moderating. For example,
P15 (male, 23, student, 3 months, 1000+) said:
I used to delete my chat history before streaming. But I could forget it. It’s not something that you
have to always keep in mind. Once I was chatting with my girlfriend about what she was doing last
night. I forgot to delete it and my audience saw that. They were making fun of it like ‘stop PDA (Public
Display of Aection).’ I don’t like it because I am supposed to be a gaming streamer, not a romantic
streamer. The content here should only be about games.
6 DISCUSSION
In this paper, we investigated streamers’ privacy concerns, strategies and challenges regarding
their information disclosure during live streaming. Through an in-depth interview study with 20
streamers, we found how streamers dened their privacy concerns in live streaming, the rationale
and dynamics behind their privacy attitudes and practices, and the challenges in managing privacy
in live streaming. Our ndings showed that while streamers were generally concerned with the
exposure of their personal information and the interruptions to their privacy boundaries with the
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
“I Am Concerned, But. . . ”: Streamers’ Privacy Concerns and Strategies In Live Streaming Information Disclosure 379:19
audience, certain information disclosure would greatly improve the attractiveness of their live
performance and increase their nancial revenues from live streaming. Thus, their privacy concerns
interplayed with their desire to maintain physical, task and social attractiveness in live streaming,
making it challenging for them to manage their information disclosure. To cope with their privacy
concerns, streamers relied on both platform-supported and self-developed privacy management
strategies, such as selective disclosure, hardware control and communicative strategies. While
these strategies cast specic outcomes on mitigating streamers’ privacy concerns, some of them
had usability issues and introduced cognitive burdens, thus needing to be further improved in the
future. We discuss the implications and contributions of these ndings in the next few subsections.
6.1 Privacy Management in Performance-Centric Information Disclosure
Our ndings reveal the intertwined relationships between streamers’ performance creation, rev-
enue seeking, and privacy protection in live streaming. Prior privacy research mostly examined
users’ privacy concerns and strategies on social media, such as Facebook and Instagram, which
are designed for online social networking, rather than for performance creation and revenue seek-
ing. These social media platforms have been mostly used for keeping in touch with others [
68
],
maintaining relationships [
22
], and reinforcing social interactions [
45
]. Thus, privacy concerns on
these platforms are mostly about how exposure of personal information [
27
] and the collapse of
dierent types of social contexts [
62
] disrupt their desired online self-presentation, online social
interactions and social relationships with dierent online social networks [1,6,44,83].
Unlike the socialization-centric social media platforms, live streaming is for users to share an
attractive performance with a vast public audience, rather than to maintain social networks. One
primary motive behind sharing an attractive performance is to make nancial revenues through
gifting from the audience [
47
]. Thus, streamers’ privacy concern is not only about exposure of
personal information, but also intertwined with whether information sharing in live streaming
enhances or undermines streamers’ performance and attractiveness to public audience, which
ultimately impacts their nancial revenue.
This socioeconomic aspect is rare in privacy issues on socialization-centric social media. It
indicates that the personal information that is appropriate or inappropriate to share is dierently
dened between live streaming and traditional socialization-centric social media. Certain personal
information, such as physical-appearance-based information, which is usually shared through
photos and generally considered to be private and to cause various privacy risks on socialization-
centric social media [
1
,
70
], can make streamers’ performance attractive to public audience. Thus, it
is a must-share for many streamers in live streaming because of its essentiality in building a personal
and attractive performance. Several interviewees, such as P2 and P5, mentioned that they chose to
be on camera to attract an audience, although they had deep concerns that exposing real physical
appearance in live streaming might reveal their identiable information. Meanwhile, some personal
information that is deemed appropriate to be shared on socialization-centric social media, such
as romantic relationships [
4
,
91
] and emotional expression [
17
,
41
], might no longer be benecial
to be shared in live streaming because, based on our interview ndings, such information might
hurt a streamers’ performance and attractiveness. For instance, P20 tried to hide her relationship
status in live streaming because it would reduce the audience’s interest in watching her streams.
Therefore, the performance-centric and the socioeconomic aspects are deeply intertwined with
the privacy issues for streamers and fundamentally inuence streamers’ privacy behaviors. The
importance of performance creation and revenue seeking should be leveraged when we evaluate
the information sharing practice and privacy concerns in the context of live streaming.
Moreover, the dynamics between streamers’ performance creation, revenue seeking, and privacy
protection entails that privacy management strategies that are eective on socialization-centric
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
379:20 Yanlai Wu, Yao Li, and Xinning Gui
social media will not necessarily function on live streaming. For example, blocking unwanted
contacts and withdrawing unwanted information, which are commonly used on socialization-centric
social media [
94
], will not work on live streaming because they reduce streamers’ attractiveness,
push the audience away, discourage audience to gift, and ultimately undermine streamers’ revenue.
As shown in our interview ndings, participants reported heavy reliance on strategies that could
both decorate their performance and preserve their privacy, such as removing unwanted content,
using beauty lters, and lying, as these strategies would not oend the audience and could potentially
increase their income. This indicates that the privacy management strategies that commonly exist
on traditional socialization-centric social media might not be sucient for streamers to protect
their privacy in live streaming. The performance-centric and the socioeconomic aspects should be
paid attention to when we design the privacy support on live streaming.
This important role of performance creation and revenue seeking in streamers’ privacy concerns
and strategies emerged from our interviews and resonates with prior work asserting streamers’
performance is a central element in live streaming and performance is driven by monetary incentives.
For example, Li et al.’s work showed how streamers prepare and perform strategic, supplementary,
and causal information presentation to create attractive performance [
48
]. Fietkiewicz et al.’s work
elaborated how streamers chose topics for their performance by evaluating the possibility to make
a living from it [
24
]. Woodcock and Johnson’s work also reported the aective labor interwoven in
the activity of live streaming, namely how streamers create content, prepare their mood, control
their stress, respond to an audience in a socially active way, and perform the identities that are
accepted by the audience [
99
]. These studies demonstrated the pervading performance-driven
and socioeconomic nature in the live streaming culture, which in our study is also deeply rooted
in streamers’ privacy concerns and strategies. We suggest future work should incorporate such
socioeconomic consideration when examining streamers’ attitudes and behaviors in the context of
live streaming.
6.2 Heightened Privacy Challenges in Compliance with Contextual Norms
Our study uncovered the heightened privacy challenges to boundary regulation due to the real-time
nature of live streaming. Prior work has touched upon how the privacy issues are worsened on live
streaming. Li et al.’s work showed that managing the disclosure of numerous pieces of information
(i.e., audio and video disclosure) at the same time while simultaneously performing (i.e., playing
games and chatting with teams) induced a signicant cognitive load on users in the streams [
50
].
However, this prior work has only examined the bystanders’ privacy issues, namely the people
being streamed by the streamers. Streamers’ privacy issues have been left unexplored. Our present
study lled this gap by focusing on the role of streamers in live streaming privacy challenges
and furthermore highlighted the severeness of the physical/virtual boundary issues in the use of
synchronous social media.
In oine communication, people’s boundary regulation is governed by context-specic informa-
tion norms [
66
]. Information sharing that is appropriate in one context may become inappropriate
in another context [
66
]. For example, when people see the social relationship, they are speaking to
changes from close to weak ties, certain information disclosure, such as intimate details and family
matters, might disrupt the interpersonal boundary. In this case, people can actively accommodate
their disclosure to the changing context because the cue of context change is visible to them.
When it comes to asynchronous social media, due to the absence of knowledge about the context
change, users can only imagine the online audience with whom they are interacting with [
62
]. The
imagined audience can be entirely dierent from the actual readers of the disclosed information
[
62
]. Thus, many researchers have proposed solutions to address the issue of imagined audience for
users to easily regulate their privacy boundaries, such as presenting users the view that a particular
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
“I Am Concerned, But. . . ”: Streamers’ Privacy Concerns and Strategies In Live Streaming Information Disclosure 379:21
audience has of their disclosed information [
51
] and informing users about the potential audience
to their information disclosure [89].
However, these solutions assume that users have sucient time to check whether their infor-
mation disclosure complies with the current context and whether they have a chance to change
their information disclosure before sharing, which are unlikely on synchronous social media.
First, streamers have little time to think about the change of context, the change of information
norms, and the compliance between their information disclosure and the norms, thus are unable
to accommodate their disclosure accordingly. For example, several participants (i.e., P10, P19 and
P7) shared their experiences where they were conducting common daily operations (i.e., logging
into an account, checking friends’ status, and browsing emails/messages), but failed to realize the
change in the context (from daily private life to live streaming) and the mismatch between their
information disclosure and the information norms on live streaming.
Second, even when streamers are able to detect the context change, they have little time to
generate an accurate estimation of the audience they are interacting with. This is because most live
streaming channels are open to anonymous public viewers, meaning anyone can join at any time
during the streams. When streamers have many anonymous viewers, it is dicult to precisely gauge
who is watching. For instance, P2 was concerned that her streams might be viewed by the subjects
she mentioned in her streams. P18 was worried that her real-world social relationships might
accidentally join her streams and view her unwanted content. Therefore, compared to asynchronous
social media, the real-time nature of synchronous social media, such as live streaming, makes
it more challenging for users to achieve contextual integrity and desired boundary regulation.
It could potentially cause more severe privacy risks, such as disclosing account credentials and
exposing identiable information. We suggest any future work pay attention to privacy challenges,
contextual integrity, and boundary regulation caused by real-time information sharing. More
research is needed to address the privacy issues in the real-time setting, since live information
sharing has become increasingly popular in recent years.
6.3 Discrimination against Female Streamers in Live Streaming
Our paper also contributed to the understanding of discrimination against female streamers in
live streaming. Prior work has shown that female streamers are generally considered to be a
marginalized and vulnerable user group in the live streaming community, since they face two major
challenges in their live streaming performance: 1) the misogynist attitudes that place women’s
bodies in opposition to the gaming culture and neglect women’s embodied presence [
72
]; 2) the
objectication that equates a woman’s worth with her body’s appearance [
65
]. Our ndings support
these prior studies, as we also witness an evident theme about the gender-based discrimination
and harassment that disturbed female streamers’ information sharing.
What we add to the prior work is that our ndings shed light on the gender issues in live
streaming from the perspective of privacy management – namely how the gender issues impede
female streamers’ information sharing and privacy management. Our study revealed that the
tension between privacy and attractiveness is a factor that makes it dicult for female streamers
to cope with the gender-based discrimination in live streaming. Prior work has shown that female
streamers suer from sexual harassment (i.e., sexual comments and solicitation, etc.), hate speech,
doxing, and appearance attack during live streaming [
72
,
88
]. Uttarapong et al.’s work found that
some viewers identied and harassed the female streamer oine based on the personal information
shared by the female streamer in streams [
88
]. Thus, the female streamers would keep personal
details minimal and avoid face cam on stream [
88
]. However, our study shows that it is not that
simple. Withholding personal details and physical features would hurt streamers’ attractiveness.
Female streamers, especially novice female streamers, need attractiveness to jumpstart their live
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
379:22 Yanlai Wu, Yao Li, and Xinning Gui
streaming career. They can be easily persuaded to trade privacy for attractiveness, making them
more vulnerable to the long-term privacy risks which they might not be able to foresee in the
moment of live streaming, such as misuse of their physical features, stalking and harassment.
For instance, P2 shared how eective it was to attract an audience when female streamers chose
to disclose their looks in their streams. P17 also mentioned how disclosing physical features
more positively inuenced attractiveness for female streamers than for male streamers. From the
perspective of privacy decision-making, users weigh gains and cost in information disclosure,
which is referred to as privacy calculus [
18
]. If users evaluate that the disclosure brings them more
benets than costs, they are highly likely to make the disclosure. In the context of live streaming,
the diculties in making rational privacy calculus under the broader discriminatory climate in
live streaming is much more for female than male streamers. The immediate benets of displaying
physical features on live streaming are unfairly amplied for female streamers. They are likely to
overestimate the immediate gain from disclosing physical features and underestimate the value of
their privacy. Thus, to facilitate female streamers to cope with the gender-based discrimination in
live streaming, both privacy protection and attractiveness management should be considered in
future research.
Second, even when female streamers value their privacy, the gender-biased discrimination on
live streaming weakens female streamers’ right to disclose or withdraw their personal information
as they wish. The penalty from the withdrawal of personal information is much more signicant
for female streamers than for male streamers, making it dicult for female streamers to decline
unwanted information sharing, such as physical appearance and age. For instance, despite of her
concern about sharing looks, P5 had suered from low attractiveness and low revenues when she
declined to be on camera. On the contrary, most male interviewees did not report such signicant
loss of attractiveness when they chose to be o camera. The unequal penalty from the withdrawal of
personal information hurts female streamers’ right to be private and disturbs their desired privacy
boundaries with audience.
Additionally, the ageism, lookism and discrimination against women with a partner that per-
vades the live streaming culture limits the ways for female streamers to freely present themselves.
Many participants mentioned in the interviews that the audience commonly expected female
streamers to be young, good-looking and single (P5, P20 and so on) and would gift such female
streamers more often. If female streamers were not so, the audience would exhibit a discriminatory
attitude, such as joking about streamers’ partners, calling streamers “aunt” and making fun of
the streamers’ looks. Such toxic expectation and attitude deny female streamers’ right to freely
self-disclose without being subject to insult or discrimination, dictate the stereotype that female
streamers can only present themselves in certain ways, and eventually impede their construction
of desired self-identities. Female streamers thus need to invest more eort than male streamers in
order to manage their audience expectation.
Our work unpacks the severe gender issues in live streaming from the perspective of privacy
management. Together with prior work that investigates the gender-based discrimination and
harassment in live streaming from other perspectives [
72
,
87
,
88
], it is obvious that gender issues
have become a common problem that aects the diversity and freedom in the live streaming
community. As live streaming has become a popular platform for people to present novel and
interesting performance, we call for more research and attention to the gender-based discriminatory
climate in live within the culture.
7 DESIGN IMPLICATIONS
Building upon our discussions of the privacy concerns and strategies that the streamers have, we
oer several design implications for live streaming platforms. To address the tension between
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
“I Am Concerned, But. . . ”: Streamers’ Privacy Concerns and Strategies In Live Streaming Information Disclosure 379:23
creating attractive performance and preserving privacy, live streaming tools should incorporate
more features that can assist streamers to make informed decisions on information sharing. Based
on our ndings, the design of these features needs to consider streamers’ privacy needs and desire
to generate attractive content in live streaming. First, to avoid streamers irrationally overestimating
the benets of disclosing personal information and underestimating the privacy risks in the long
run, the live streaming tools should provide sucient information to raise streamers’ awareness of
the privacy risks associated with their information disclosure. In our interviews, it was common
that streamers, especially female streamers, were well aware of the attractiveness obtained from
information sharing and underestimated the associated long-term privacy risk, such as being
identied oine. To raise streamers’ awareness of the long-term privacy risks, an eective approach
is to provide streamers with online interactive tutorials that introduce possible privacy consequences
resulting from overexposure of personal information in live streaming and coping strategies. The
tutorials can be displayed on the rst-time streamers live broadcast and periodically. Another
example is to highlight true stories about privacy breaches that have happened to other streamers
in the news feed. A few participants, such as P10 and P11, mentioned that the true stories they
encountered online alerted them about the privacy risks from live streaming, warned them to be
more careful about their personal information, and thus helped them make better privacy decisions.
Thus, incorporating and highlighting such information can eectively improve streamers’ privacy
awareness.
Second, live streaming tools should provide features for streamers to hide unwanted information
disclosure but not hurt their attractiveness. Current features that are commonly used by streamers
to hide unwanted information disclosure include webcam lters and overlays. The former can be
used to hide streamers’ physical features and the latter can be used to hide screensharing content.
However, these features do not consider streamers’ desire to maintain attractiveness and retain
audience members because they disrupt the streamers’ performance (mentioned by P20) and make
it obvious that streamers want to keep something secret from the audience (mentioned by P2).
Therefore, live streaming tools should oer features that take both attractiveness and privacy into
account. For example, webcam lters and overlays can be more intelligent and adaptive so that they
can be automatically and naturally accommodated to the streamed performance. Live streaming
tools should provide features for streamers to better understand an audience’s need and evaluate the
eect of their privacy management strategies so that streamers can develop diverse approaches to
attract audience members, instead of having to disclose unwanted personal information. Examples
can be embedded surveys and interviews in the live streams. Such features can also give streamers
more information and time to better decide the trade-o between performance creation and privacy
protection.
Intelligent notication features and mechanisms that can assist streamers to notice the real-
time change of information sharing context should be oered in live streaming tools. Notication
of change of context is not new. Many live streaming tools already provide notications when
new audience members join in or when a recording is started. However, such notications are
not sucient because, based on our interview ndings, many changes of context, such as a pop-
up login page, calls from real-world connections, push messages, in-game communications, and
bystanders’ interruptions, are not included in notications from the live streaming tools. Streamers
have to perform extra tasks, such as disabling notications in the device settings, adjusting the
audio/video sources and adding overlays, to address dierent changes of context. However, these
extra tasks indiscriminately block information under all the contexts and require much labor from
the streamers.
We thus suggest the notication of change of context be more integrated, intelligent, and
granular. First, the notication of change of context should be congured in live streaming platforms
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
379:24 Yanlai Wu, Yao Li, and Xinning Gui
and integrated with other software in the live broadcast (i.e., games, audio chat, web browser, etc.).
Streamers can perform the congurations of notications in one place without having to adjust
settings in other software. Second, the notication of change of context should be more intelligent,
adaptive and predictive. For example, the notication should be able to detect an incoming pop-
up login page, calls from real-world connections, push messages, in-game communications, and
bystanders’ interruptions during live streaming, and predict the appropriateness based on the
current audience in the live stream. Thus, streamers can decide whether to share information under
the changed context before being streamed. This might require context-awareness and intelligent
techniques to support the implementation of the notications. Lastly, the notication content
should be granular so that the streamers can have more information about the change of context to
make informed privacy decisions. For instance, dierent colors can be used to show the appropriate
levels of the change of context. A summary of the change of context and the information about the
current audience should be included in the notications. Streamers can customize the notication
content and format to help them be aware of the change of context.
Lastly, to address gender issues in the privacy management in live streaming, moderations to
the discriminatory attitudes and behaviors that make female streamers vulnerable to privacy risks
should be given sucient attention in the design of live streaming tools. There has been increased
discussion on developing moderation practices in the live streaming community [
9
,
10
,
96
]. Many
live streaming platforms have well-trained employees to handle inappropriate broadcasting content.
Streamers either moderate the audience’s comments on their own or hire volunteer moderators
[
75
,
76
,
96
]. There also exist automatic moderation tools that use algorithms to lter abusive
messages. However, most of these moderation mechanisms focus on harassment, trolls, and hate
speech in live streaming. Few of them focus on privacy issues or are specic to the gender-based
discrimination. Thus, moderation mechanisms that are designed for the privacy and gender issues
in live streaming are needed. We suggest more moderators from marginalized populations and
moderators who have knowledge about privacy management should be included in the moderator
community in live streaming, as they may understand female streamers and other marginalized
streamers’ struggles and privacy challenges better. Previous work points out that users generally
preferred friends to be moderators compared to paid or volunteer strangers to protect personal
information [
60
]. However, turning for help to friends may not work for female streamers in live
streaming because it might break the interpersonal boundary. P18, for example, did not want her
friends to be aware of her broadcasting. Therefore, female streamers may choose loyal female
audience or peer female streamers to be their moderators who have long-term interaction with
the streamers and understand their gender-specic concerns and needs. Additionally, as Zolides
has studied how the community guidelines and policies structurally encourage forms of gendered
domination by reinforcing hegemonic and disempowering norms of femininity in an already
highly masculine space [
104
], we call for more policy research and design to address the harmful
understandings of gender inuenced by the content moderation policies and community guidelines.
8 LIMITATIONS AND FUTURE WORK
Our study has several limitations. First, all the participants are Chinese DouYu streamers. As such,
our results may not be generalized to users from other countries or cultural backgrounds. Prior
research on cross-cultural studies has shown that people’s privacy concerns and strategies toward
information disclosure on social media could be distinct from culture to culture [
3
,
38
]. Given that
all our participants are Chinese, their privacy concerns and strategies in live streaming might be
inuenced by the Chinese culture. For instance, due to the salient importance of social relationships
and social identities in the Chinese culture, information sharing that is socially unacceptable might
particularly concern the Chinese streamers. However, such post hoc speculations based on the prior
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
“I Am Concerned, But. . . ”: Streamers’ Privacy Concerns and Strategies In Live Streaming Information Disclosure 379:25
literature need to be further examined in future studies with specic culture-related probes. The
primary goal of the present study is to explore streamers’ privacy concerns and strategies on a novel
social media platform. Hence, we didn’t include cultural related questions in the interviews. Future
work is suggested to investigate the role of cultural factors in streamers’ information disclosure
management.
Second, the majority of our participants (17 out of 20) stream about games, with a few exceptions
who stream about singing, dancing, and so on (3 out of 20). This is because DouYu is a leading
game-centric live streaming platform in China [
20
] and gameplay streaming is a popular live
streaming scenario which is also the case for other platforms [
64
]. However, it is possible that
streamers who stream other topics, such as travel experience, shopping, and cooking, may have
dierent privacy attitudes and behaviors. Hence, we suggest future work explore streamers’ privacy
management who stream topics other than what we have covered in this paper.
Third, although we recruited through diverse channels, our participants are mostly well-
educated. This might be because more than 65% streamers on DouYu have a college educational
background [
33
]. Streamers who are less educated may have dierent privacy concerns from the
well-educated streamers. Therefore, our sample may not well represent streamers with dierent
education backgrounds. We suggest future work explore the privacy concerns and strategies of
streamers with diverse educational backgrounds.
Fourth, although we considered full-time/part-time, length of streaming experience and number
of viewers when recruiting streamers, we failed to interview more full-time streamers and streamers
who were famous, had hundreds of thousands of viewers, and developed their streaming into a
successful career. This is because most famous streamers usually have contracts with DouYu or other
agencies, which pose restrictions on streamers being interviewed. These streamers may encounter
dierent privacy issues and adopt dierent strategies to manage their information disclosure.
Therefore, we suggest future work develop better strategies to recruit such streamers. We also
suggest large-scale quantitative studies examine whether professionalism, length of streaming
experience and number of viewers would signicantly inuence streamers’ privacy attitudes and
behaviors.
9 CONCLUSION
In this paper, we presented an investigation of streamers’ privacy concerns and strategies in live
streaming through an interview study with 20 streamers. We found that streamers were concerned
with the information disclosure that would harm their attractiveness to the audience and disturb
their interpersonal boundary with the audience. To cope with their privacy concerns, streamers
relied on both platform-supported and self-developed privacy management strategies, such as
selective disclosure, hardware control and communicative strategies. While these strategies cast a
certain eect on mitigating streamers’ privacy concerns, some of them had usability problems and
required too much cognitive eort, thus needing to be further improved in the future. We discuss
empirical contributions and design implications of our ndings.
REFERENCES
[1]
Shane Ahern, Dean Eckles, Nathaniel S Good, Simon King, Mor Naaman, and Rahul Nair. 2007. Over-exposed?
Privacy patterns and considerations in online and mobile photo sharing. In Proceedings of the SIGCHI conference on
Human factors in computing systems. 357–366.
[2] Irwin Altman. 1976. A conceptual analysis. Environment and behavior 8, 1 (1976), 7–29.
[3]
Irwin Altman. 1977. Privacy regulation: Culturally universal or culturally specic? Journal of social issues 33, 3 (1977),
66–84.
[4]
Leslie A Baxter and Sally Widenmann. 1993. Revealing and not revealing the status of romantic relationships to
social networks. Journal of Social and Personal Relationships 10, 3 (1993), 321–337.
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
379:26 Yanlai Wu, Yao Li, and Xinning Gui
[5] Anchor Books and Erving Goman. 1969. The presentation of self in everyday life. London: Allen Lane (1969).
[6]
Danah Boyd. 2008. Why youth (heart) social network sites: The role of networked publics in teenage social life.
YOUTH, IDENTITY, AND DIGITAL MEDIA, David Buckingham, ed., The John D. and Catherine T. MacArthur Foundation
Series on Digital Media and Learning, The MIT Press, Cambridge, MA (2008), 2007–16.
[7]
Virginia Braun and Victoria Clarke. 2006. Using thematic analysis in psychology. Qualitative research in psychology 3,
2 (2006), 77–101.
[8]
Simon Bründl and Thomas Hess. 2016. Why do users broadcast? Examining individual motives and social capital on
social live streaming platforms. (2016).
[9]
Jie Cai and Donghee Yvette Wohn. 2019. Categorizing live streaming moderation tools: An analysis of twitch.
International Journal of Interactive Communication Systems and Technologies (IJICST) 9, 2 (2019), 36–50.
[10]
Jie Cai and Donghee Yvette Wohn. 2021. After Violation But Before Sanction: Understanding Volunteer Moderators’
Proling Processes Toward Violators in Live Streaming Communities. Proceedings of the ACM on Human-Computer
Interaction 5, CSCW2 (2021), 1–25.
[11]
Matthew Carrasco and Andruid Kerne. 2018. Queer visibility: Supporting LGBT+ selective visibility on social media.
In Proceedings of the 2018 CHI conference on human factors in computing systems. 1–12.
[12]
Chia-Chen Chen and Yi-Chen Lin. 2018. What drives live-stream usage intention? The perspectives of ow, enter-
tainment, social interaction, and endorsement. Telematics and Informatics 35, 1 (2018), 293–303.
[13]
Yue Chen, Qin Gao, and Pei-Luen Patrick Rau. 2017. Watching a movie alone yet together: understanding reasons for
watching Danmaku videos. International Journal of Human–Computer Interaction 33, 9 (2017), 731–743.
[14]
Hichang Cho and Anna Filippova. 2016. Networked privacy management in Facebook: A mixed-methods and
multinational study. In Proceedings of the 19th ACM Conference on Computer-Supported Cooperative Work & Social
Computing. 503–514.
[15] J. Clement. 2022. Twitch - statistics & facts. https://www.statista.com/topics/7946/twitch/.
[16]
Sauvik Das and Adam Kramer. 2013. Self-censorship on Facebook. In Proceedings of the International AAAI Conference
on Web and Social Media, Vol. 7. 120–127.
[17]
Daantje Derks, Agneta H Fischer, and Arjan ER Bos. 2008. The role of emotion in computer-mediated communication:
A review. Computers in human behavior 24, 3 (2008), 766–785.
[18]
Tamara Dinev and Paul Hart. 2006. An extended privacy calculus model for e-commerce transactions. Information
systems research 17, 1 (2006), 61–80.
[19]
Audubon Dougherty. 2011. Live-streaming mobile video: production as civic engagement. In Proceedings of the 13th
international conference on human computer interaction with mobile devices and services. 425–434.
[20]
DouYu International Holdings Limited Reports First Quarter 2021 Unaudited Financial Results. 2021. https://ir.douyu.
com/Press-Releases?l=10.
[21]
Nicole Ellison, Rebecca Heino, and Jennifer Gibbs. 2006. Managing impressions online: Self-presentation processes in
the online dating environment. Journal of computer-mediated communication 11, 2 (2006), 415–441.
[22]
Nicole B Ellison, Charles Steineld, and Cli Lampe. 2007. The benets of Facebook “friends:” Social capital and college
students’ use of online social network sites. Journal of computer-mediated communication 12, 4 (2007), 1143–1168.
[23]
Encinitas home of famous online gamer “DrDisRespect” shot at with BB gun. 2018. https://www.sandiegouniontribune.
com/news/public-safety/sd-me-twitch- shooting-20180911- story.html.
[24]
Kaja J Fietkiewicz, Isabelle Dorsch, Katrin Scheibe, Franziska Zimmer, and Wolfgang G Stock. 2018. Dreaming of
stardom and money: Micro-celebrities and inuencers on live streaming services. In International Conference on Social
Computing and Social Media. Springer, 240–253.
[25]
Guo Freeman and Donghee Yvette Wohn. 2020. Streaming your identity: Navigating the presentation of gender and
sexuality through live streaming. Computer Supported Cooperative Work (CSCW) 29, 6 (2020), 795–825.
[26]
Reesha Gandhi, Christine L Cook, Nina LaMastra, Jirassaya Uttarapong, and Donghee Yvette Wohn. 2021. An
exploration of mental health discussions in live streaming gaming communities. Frontiers in psychology 12 (2021),
575653.
[27]
Ralph Gross and Alessandro Acquisti. 2005. Information revelation and privacy in online social networks. In
Proceedings of the 2005 ACM workshop on Privacy in the electronic society. 71–80.
[28]
Gunman targets YouTubers in home invasion. 2018. https://web.archive.org/web/20180213235353/https://www.
theverge.com/2018/2/13/17009026/youtube-vloggers-gunman-invasion.
[29]
Oliver L Haimson and John C Tang. 2017. What makes live events engaging on Facebook Live, Periscope, and
Snapchat. In Proceedings of the 2017 CHI conference on human factors in computing systems. 48–60.
[30]
William A Hamilton, Oliver Garretson, and Andruid Kerne. 2014. Streaming on twitch: fostering participatory
communities of play within live mixed media. In Proceedings of the SIGCHI conference on human factors in computing
systems. 1315–1324.
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
“I Am Concerned, But. . . ”: Streamers’ Privacy Concerns and Strategies In Live Streaming Information Disclosure 379:27
[31]
Rakibul Hasan, Patrick Shaer, David Crandall, Eman T Apu Kapadia, et al
.
2017. Cartooning for enhanced privacy in
lifelogging and streaming videos. In Proceedings of the IEEE conference on computer vision and pattern recognition
workshops. 29–38.
[32]
Zorah Hilvert-Bruce, James T Neill, Max Sjöblom, and Juho Hamari. 2018. Social motivations of live-streaming viewer
engagement on Twitch. Computers in Human Behavior 84 (2018), 58–67.
[33]
How Do You Think: Over 60% of Live Streamers Have A Bachelor’s Degree; 80% of Young People In New Business
Forms Earn Less Than 8,000 RMB Per Month? (Translated Title). 2021. https://www.sohu.com/a/www.sohu.com/a/
511712243_507132.
[34]
Mu Hu, Mingli Zhang, and Yu Wang. 2017. Why do audiences choose to keep watching on live video streaming
platforms? An explanation of dual identication framework. Computers in Human Behavior 75 (2017), 594–606.
[35]
Adele Lu Jia, Siqi Shen, Dick HJ Epema, and Alexandru Iosup. 2016. When game becomes life: The creators and
spectators of online game replays and live streaming. ACM Transactions on Multimedia Computing, Communications,
and Applications (TOMM) 12, 4 (2016), 1–24.
[36]
Oskar Juhlin, Arvid Engström, and Erika Reponen. 2010. Mobile broadcasting: the whats and hows of live video as a
social medium. In Proceedings of the 12th international conference on Human computer interaction with mobile devices
and services. 35–44.
[37]
Sanjay Kairam, Mike Brzozowski, David Huaker, and Ed Chi. 2012. Talking in circles: selective sharing in google+.
In Proceedings of the SIGCHI conference on human factors in computing systems. 1065–1074.
[38]
Yoojung Kim, Dongyoung Sohn, and Sejung Marina Choi. 2011. Cultural dierence in motivations for using social
network sites: A comparative study of American and Korean college students. Computers in human behavior 27, 1
(2011), 365–372.
[39]
Christoph Klimmt, Tilo Hartmann, and Holger Schramm. 2006. Parasocial interactions and relationships. Psychology
of entertainment (2006), 291–313.
[40]
Johannes Knoll, Holger Schramm, Christiana Schallhorn, and Sabrina Wynistorf. 2015. Good guy vs. bad guy: The
inuence of parasocial interactions with media characters on brand placement eects. International Journal of
Advertising 34, 5 (2015), 720–743.
[41]
Adam DI Kramer. 2012. The spread of emotion via Facebook. In Proceedings of the SIGCHI conference on human factors
in computing systems. 767–770.
[42]
Kate Szer Kurtin, Nina O’Brien, Deya Roy, and Linda Dam. 2018. The development of parasocial interaction
relationships on YouTube. The Journal of Social Media in Society 7, 1 (2018), 233–252.
[43]
Nina LaMastra, Jirassaya Uttarapong, Reesha Gandhi, Christine L Cook, and Donghee Yvette Wohn. 2020. How a Live
Streamer’s Choice in Played Game Aects Mental Health Conversations. In Extended Abstracts of the 2020 Annual
Symposium on Computer-Human Interaction in Play. 297–300.
[44]
Airi Lampinen, Vilma Lehtinen, Asko Lehmuskallio, and Sakari Tamminen. 2011. We’re in it together: interpersonal
management of disclosure in social network services. In Proceedings of the SIGCHI conference on human factors in
computing systems. 3217–3226.
[45]
Eunji Lee, Jung-Ah Lee, Jang Ho Moon, and Yongjun Sung. 2015. Pictures speak louder than words: Motivations for
using Instagram. Cyberpsychology, behavior, and social networking 18, 9 (2015), 552–556.
[46]
Yu-Hao Lee, Chien Wen Yuan, and Donghee Yvette Wohn. 2021. How video streamers’ mental health disclosures
aect viewers’ risk perceptions. Health Communication 36, 14 (2021), 1931–1941.
[47]
Jie Li, Xinning Gui, Yubo Kou, and Yukun Li. 2019. Live streaming as co-performance: Dynamics between center and
periphery in theatrical engagement. Proceedings of the ACM on human-computer interaction 3, CSCW (2019), 1–22.
[48]
Lingyuan Li, Jirassaya Uttarapong, Guo Freeman, and Donghee Yvette Wohn. 2020. Spontaneous, Yet Studious: Esports
Commentators’ Live Performance and Self-Presentation Practices. Proceedings of the ACM on Human-Computer
Interaction 4, CSCW2 (2020), 1–25.
[49]
Yao Li, Xinning Gui, Yunan Chen, Heng Xu, and Alfred Kobsa. 2018. When SNS privacy settings become granular:
investigating users’ choices, rationales, and inuences on their social experience. Proceedings of the ACM on Human-
Computer Interaction 2, CSCW (2018), 1–21.
[50]
Yao Li, Yubo Kou, Je Seok Lee, and Alfred Kobsa. 2018. Tell me before you stream me: Managing information disclosure
in video game live streaming. Proceedings of the ACM on Human-Computer Interaction 2, CSCW (2018), 1–18.
[51]
Heather Richter Lipford, Jason Watson, Michael Whitney, Katherine Froiland, and Robert W Reeder. 2010. Visual vs.
compact: A comparison of privacy policy interfaces. In Proceedings of the SIGCHI Conference on Human Factors in
Computing Systems. 1111–1114.
[52]
Xing Lu and Zhicong Lu. 2019. Fifteen seconds of fame: A qualitative study of Douyin, a short video sharing mobile
application in China. In International Conference on human-computer interaction. Springer, 233–244.
[53]
Zhicong Lu. 2019. Live streaming in China for sharing knowledge and promoting intangible cultural heritage.
Interactions 27, 1 (2019), 58–63.
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
379:28 Yanlai Wu, Yao Li, and Xinning Gui
[54]
Zhicong Lu, Michelle Annett, Mingming Fan, and Daniel Wigdor. 2019. " I feel it is my responsibility to stream"
Streaming and Engaging with Intangible Cultural Heritage through Livestreaming. In Proceedings of the 2019 CHI
Conference on Human Factors in Computing Systems. 1–14.
[55]
Zhicong Lu, Michelle Annett, and Daniel Wigdor. 2019. Vicariously experiencing it all without going outside: A study
of outdoor livestreaming in China. Proceedings of the ACM on Human-Computer Interaction 3, CSCW (2019), 1–28.
[56]
Zhicong Lu, Chenxinran Shen, Jiannan Li, Hong Shen, and Daniel Wigdor. 2021. More kawaii than a real-person live
streamer: understanding how the otaku community engages with and perceives virtual YouTubers. In Proceedings of
the 2021 CHI Conference on Human Factors in Computing Systems. 1–14.
[57]
Zhicong Lu, Haijun Xia, Seongkook Heo, and Daniel Wigdor. 2018. You watch, you give, and you engage: a study of
live streaming practices in China. In Proceedings of the 2018 CHI conference on human factors in computing systems.
1–13.
[58]
Courtney Rees Lyles, Lynne T Harris, Tung Le, Jan Flowers, James Tufano, Diane Britt, James Hoath, Irl B Hirsch,
Harold I Goldberg, and James D Ralston. 2011. Qualitative evaluation of a mobile phone and web-based collaborative
care intervention for patients with type 2 diabetes. Diabetes technology & therapeutics 13, 5 (2011), 563–569.
[59] Jerey K MacKie-Mason. 2011. All Space Will Be Public Space. (2011).
[60]
Kaitlin Mahar, Amy X Zhang, and David Karger. 2018. Squadbox: A tool to combat email harassment using friend-
sourced moderation. In Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems. 1–13.
[61]
Alice E Marwick. 2005. ’I’m a Lot More Interesting than a Friendster Prole’: Identity Presentation, Authenticity and
Power in Social Networking Services. Association of Internet Researchers 6 (2005).
[62]
Alice E Marwick and Danah Boyd. 2011. I tweet honestly, I tweet passionately: Twitter users, context collapse, and
the imagined audience. New media & society 13, 1 (2011), 114–133.
[63] James C McCroskey and Thomas A McCain. 1974. The measurement of interpersonal attraction. (1974).
[64] Most Watched Games on Twitch. 2021. https://twitchtracker.com/games.
[65]
Supun Nakandala, Giovanni Ciampaglia, Norman Su, and Yong-Yeol Ahn. 2017. Gendered conversation in a social
game-streaming platform. In Proceedings of the international AAAI conference on web and social media, Vol. 11. 162–171.
[66] Helen Nissenbaum. 2009. Privacy in context. In Privacy in Context. Stanford University Press.
[67]
Xinru Page, Alfred Kobsa, and Bart Knijnenburg. 2012. Don’t disturb my circles! Boundary preservation is at the
center of location-sharing concerns. In Proceedings of the International AAAI Conference on Web and Social Media,
Vol. 6. 266–273.
[68]
Namsu Park, Kerk F Kee, and Sebastián Valenzuela. 2009. Being immersed in social networking environment: Facebook
groups, uses and gratications, and social outcomes. Cyberpsychology & behavior 12, 6 (2009), 729–733.
[69]
Anthony J Pellicone and June Ahn. 2017. The Game of Performing Play: Understanding streaming as cultural
production. In Proceedings of the 2017 CHI conference on human factors in computing systems. 4863–4874.
[70]
Yasmeen Rashidi, Tousif Ahmed, Felicia Patel, Emily Fath, Apu Kapadia, Christena Nippert-Eng, and Norman Makoto
Su. 2018. " You don’t want to be the next meme": College Students’ Workarounds to Manage Privacy in the Era of
Pervasive Photography. In Fourteenth Symposium on Usable Privacy and Security (SOUPS 2018). 143–157.
[71]
Bonnie Ruberg and Amanda LL Cullen. 2019. Feeling for an Audience. Digital Culture & Society 5, 2 (2019), 85–102.
[72]
Bonnie Ruberg, Amanda LL Cullen, and Kathryn Brewster. 2019. Nothing but a “titty streamer”: legitimacy, labor,
and the debate over women’s breasts in video game live streaming. Critical Studies in Media Communication 36, 5
(2019), 466–481.
[73] Rebecca B Rubin and Michael P McHugh. 1987. Development of parasocial interaction relationships. (1987).
[74]
Anthony Russo. 2019. DouYu International Sees User Growth, Doubled Revenue,. https://old.capitalwatch.com/article-
4697-1.html.
[75]
Joseph Seering, Robert Kraut, and Laura Dabbish. 2017. Shaping pro and anti-social behavior on twitch through
moderation and example-setting. In Proceedings of the 2017 ACM conference on computer supported cooperative work
and social computing. 111–125.
[76]
Joseph Seering, Tony Wang, Jina Yoon, and Geo Kaufman. 2019. Moderator engagement and community development
in the age of algorithms. New Media & Society 21, 7 (2019), 1417–1443.
[77]
Pan Shi, Heng Xu, and Yunan Chen. 2013. Using contextual integrity to examine interpersonal information boundary
on social network sites. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems. 35–38.
[78]
Samarth Singhal, Carman Neustaedter, Thecla Schiphorst, Anthony Tang, Abhisekh Patra, and Rui Pan. 2016. You are
Being Watched: Bystanders’ Perspective on the Use of Camera Devices in Public Spaces. In Proceedings of the 2016
CHI Conference Extended Abstracts on Human Factors in Computing Systems. 3197–3203.
[79]
Max Sjöblom and Juho Hamari. 2017. Why do people watch others play video games? An empirical study on the
motivations of Twitch users. Computers in human behavior 75 (2017), 985–996.
[80]
Marita Skjuve and Petter Bae Brandtzaeg. 2019. Facebook live: A mixed-methods approach to explore individual live
streaming practices and motivations on Facebook. Interacting with Computers 31, 3 (2019), 589–602.
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
“I Am Concerned, But. . . ”: Streamers’ Privacy Concerns and Strategies In Live Streaming Information Disclosure 379:29
[81]
Manya Sleeper, Rebecca Balebako, Sauvik Das, Amber Lynn McConahy, Jason Wiese, and Lorrie Faith Cranor. 2013.
The post that wasn’t: exploring self-censorship on facebook. In Proceedings of the 2013 conference on Computer
supported cooperative work. 793–802.
[82]
Thomas Smith, Marianna Obrist, and Peter Wright. 2013. Live-streaming changes the (video) game. In Proceedings of
the 11th european conference on Interactive TV and video. 131–138.
[83]
Frederic Stutzman and Woodrow Hartzog. 2012. Boundary regulation in social media. In Proceedings of the ACM 2012
conference on computer supported cooperative work. 769–778.
[84]
Fred Stutzman and Jacob Kramer-Dueld. 2010. Friends only: examining a privacy-enhancing behavior in facebook.
In Proceedings of the SIGCHI conference on human factors in computing systems. 1553–1562.
[85] John C Tang, Gina Venolia, and Kori M Inkpen. 2016. Meerkat and periscope: I stream, you stream, apps stream for
live streams. In Proceedings of the 2016 CHI conference on human factors in computing systems. 4770–4780.
[86]
Patricia R Todd and Joanna Melancon. 2019. Gender dierences in perceptions of trolling in livestream video
broadcasting. Cyberpsychology, Behavior, and Social Networking 22, 7 (2019), 472–476.
[87]
Lena Uszkoreit. 2018. With great power comes great responsibility: video game live streaming and its potential risks
and benets for female gamers. In Feminism in Play. Springer, 163–181.
[88]
Jirassaya Uttarapong, Jie Cai, and Donghee Yvette Wohn. 2021. Harassment Experiences of Women and LGBTQ
Live Streamers and How They Handled Negativity. In ACM International Conference on Interactive Media Experiences.
7–19.
[89]
Yang Wang, Pedro Giovanni Leon, Alessandro Acquisti, Lorrie Faith Cranor, Alain Forget, and Norman Sadeh. 2014.
A eld trial of privacy nudges for facebook. In Proceedings of the SIGCHI conference on human factors in computing
systems. 2367–2376.
[90]
Yang Wang, Yao Li, Bryan Semaan, and Jian Tang. 2016. Space collapse: reinforcing, reconguring and enhancing
Chinese social practices through WeChat. In Tenth International AAAI Conference on Web and Social Media.
[91]
Yang Wang, Gregory Norcie, Saranga Komanduri, Alessandro Acquisti, Pedro Giovanni Leon, and Lorrie Faith Cranor.
2011. " I regretted the minute I pressed share" a qualitative study of regrets on Facebook. In Proceedings of the seventh
symposium on usable privacy and security. 1–16.
[92]
Jason Watson, Andrew Besmer, and Heather Richter Lipford. 2012. + Your circles: sharing behavior on Google+. In
Proceedings of the eighth symposium on usable privacy and security. 1–9.
[93] Alan F Westin. 1968. Privacy and freedom. Washington and Lee Law Review 25, 1 (1968), 166.
[94]
Pamela Wisniewski, Heather Lipford, and David Wilson. 2012. Fighting for my space: Coping mechanisms for SNS
boundary regulation. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems. 609–618.
[95]
Pamela J Wisniewski, Bart P Knijnenburg, and Heather Richter Lipford. 2017. Making privacy personal: Proling
social network users to inform privacy education and nudging. International Journal of human-computer studies 98
(2017), 95–108.
[96] Donghee Yvette Wohn. 2019. Volunteer moderators in twitch micro communities: How they get involved, the roles
they play, and the emotional labor they experience. In Proceedings of the 2019 CHI conference on human factors in
computing systems. 1–13.
[97]
Donghee Yvette Wohn and Guo Freeman. 2020. Audience management practices of live streamers on Twitch. In ACM
International Conference on Interactive Media Experiences. 106–116.
[98]
Donghee Yvette Wohn, Guo Freeman, and Caitlin McLaughlin. 2018. Explaining viewers’ emotional, instrumental,
and nancial support provision for live streamers. In Proceedings of the 2018 CHI conference on human factors in
computing systems. 1–13.
[99]
Jamie Woodcock and Mark R Johnson. 2019. The aective labor and performance of live streaming on Twitch. tv.
Television & New Media 20, 8 (2019), 813–823.
[100]
Josh Ye. 2019. How Douyu won the live-streaming war to become China’s Twitch. https://www.scmp.com/abacus/who-
what/what/article/3028270/how-douyu-won-live-streaming-war-become-chinas- twitch.
[101]
Ge Zhang and Larissa Hjorth. 2019. Live-streaming, games and politics of gender performance: The case of Nüzhubo
in China. Convergence 25, 5-6 (2019), 807–825.
[102]
Shanyang Zhao, Sherri Grasmuck, and Jason Martin. 2008. Identity construction on Facebook: Digital empowerment
in anchored relationships. Computers in human behavior 24, 5 (2008), 1816–1836.
[103]
Zhenhui Zhu, Zhi Yang, and Yafei Dai. 2017. Understanding the gift-sending interaction on live-streaming video
websites. In International Conference on social computing and social media. Springer, 274–285.
[104]
Andrew Zolides. 2021. Gender moderation and moderating gender: Sexual content policies in Twitch’s community
guidelines. New Media & Society 23, 10 (2021), 2999–3015.
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
379:30 Yanlai Wu, Yao Li, and Xinning Gui
10 APPENDIX
Interview Questions
Streaming Experience:
What’s your gender?
What’s your age?
What’s your occupation?
How did you know DouYu?
How long have you been streamed on DouYu?
Why did you choose to stream on DouYu?
Streaming Content:
What topic(s) did you broadcast on DouYu? Why did you choose such topic(s)?
When you broadcast the topic(s), what information would you like to share?
Could you please describe that in detail with an example?
Why did you want to share such information?
Who were the audience?
How did the audience react to your information sharing?
Did you have any concerns when you shared such information?
If yes, what concern? Why did you have such concern?
What happened after you shared the information?
What information did you prefer not to share with your audience?
What strategies did you adopt to prevent unwanted information sharing?
Have you ever shared any information that you didn’t want to share with your
audience?
What happened at that time? What information did you share?
Why did you think such information sharing was unwanted?
What did you do when the incident happened? How did the audience react?
Did the incident aect your life?
Did you have any regrets on it?
Streaming Equipment:
What equipment do you use during live streaming? (e.g., webcam, mic, etc.)
How did you use it?
Why did you use such equipment?
When you used such equipment, what information could the audience get?
What information sharing was acceptable to you? What not? Why?
How did you prevent unwanted information sharing?
If you didn’t want to use certain equipment, why not use this equipment?
How did you disable such equipment?
Viewers:
How many viewers did you have?
Who were your viewers?
Did the viewers have an impact on your streaming? How?
How did the viewers react to your streaming?
When you interacted with the viewers, have you ever accidentally shared personal
information?
Have your viewers ever asked you any unwanted questions or made any inappropri-
ate request to solicit your personal information?
What questions/requests?
Proc. ACM Hum.-Comput. Interact., Vol. 6, No. CSCW2, Article 379. Publication date: November 2022.
“I Am Concerned, But. . . ”: Streamers’ Privacy Concerns and Strategies In Live Streaming Information Disclosure 379:31
How did you respond at that time?
Usage of Privacy Setting:
What features did you often use on DouYu? What do you think of them